Place · Level 3
Melatonin
全球年销 $1.5B+ · 天然睡眠激素营销 · 但商业剂量 5-10 mg 远超有效剂量 0.3 mg · 真适应症窄
Last updated
Story path
- 1Pineal synthesis · dark signalPineal synthesis · dark signal
- 20.3 mg vs 5-10 mg · dose myth0.3 mg vs 5-10 mg · dose myth
- 3Jet lag + shift work · best indicationJet lag + shift work · best indication
- 4Label chaos + commercial issuesLabel chaos + commercial issues
- 5Stronger interventions than melatoninStronger interventions than melatonin
- 6Summary · should I take itSummary · should I take it
Chapter 1
Pineal synthesis · dark signal
Pineal synthesis · dark signal
Melatonin is the body's own 'nighttime signal' hormone, secreted mainly by the pineal gland deep in the brain.
Synthesis chain: Tryptophan → 5-hydroxytryptophan → serotonin (5-HT) → N-acetylserotonin → melatonin. The key enzyme is AANAT (aralkylamine N-acetyltransferase), regulated by the circadian clock.
The control switch is the suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. (suprachiasmatic nucleus): retinal ipRGCs (containing melanopsin) sense blue light (~480 nm). During the day the SCN suppresses AANAT and melatonin is low; in darkness the SCN releases suppression, AANAT is upregulated, and melatonin surges.
Typical rhythm:
Daytime plasma <10 pg/mL9–10 pm starts rising (Dim Light Melatonin Onset, DLMO)2–4 am peak — it varies widely between individuals (roughly tens to a couple of hundred pg/mL; higher in children, much lower in the elderly)Returns to baseline by 6–7 am
Real role in sleep chemistry
Melatonin doesn't directly make you fall asleep — it signals to the body that 'it's now night'. Sleep itself is the combined effect of adenosine, sleep pressure, and the circadian clock. Melatonin's actual job is phase-shifting the body clock, dropping core temperature 0.3–0.5°C, and weakening the wake drive.
What aging and modern life do to it
After age 30 the peak declines with age, and by old age it is markedly lower than in a young adult — but published figures differ too much to quote a single percentage, so treat this as a direction, not a number. Modern blue-light exposure (phones, computers, LEDs) pushes the same way: in a controlled trial, five evenings of e-reader use suppressed melatonin about 55% and delayed its onset (see the screens page below); ordinary evening screen use is a weaker version of the same effect. This explains why modern people have widely disrupted sleep rhythms, why the elderly struggle to fall asleep, and why blue-light glasses or phone night mode actually make sense.
One common misconception: 'more serotonin = better sleep' is only half right. Serotonin is melatonin's precursor, but supplementing serotonin or tryptophan alone won't significantly raise melatonin, because other regulatory steps gate the path.
Synthesis chain: Tryptophan → 5-hydroxytryptophan → serotonin (5-HT) → N-acetylserotonin → melatonin. The key enzyme is AANAT (aralkylamine N-acetyltransferase), regulated by the circadian clock.
The control switch is the suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. (suprachiasmatic nucleus): retinal ipRGCs (containing melanopsin) sense blue light (~480 nm). During the day the SCN suppresses AANAT and melatonin is low; in darkness the SCN releases suppression, AANAT is upregulated, and melatonin surges.
Typical rhythm:
Daytime plasma <10 pg/mL9–10 pm starts rising (Dim Light Melatonin Onset, DLMO)2–4 am peak — it varies widely between individuals (roughly tens to a couple of hundred pg/mL; higher in children, much lower in the elderly)Returns to baseline by 6–7 am
Real role in sleep chemistry
Melatonin doesn't directly make you fall asleep — it signals to the body that 'it's now night'. Sleep itself is the combined effect of adenosine, sleep pressure, and the circadian clock. Melatonin's actual job is phase-shifting the body clock, dropping core temperature 0.3–0.5°C, and weakening the wake drive.
What aging and modern life do to it
After age 30 the peak declines with age, and by old age it is markedly lower than in a young adult — but published figures differ too much to quote a single percentage, so treat this as a direction, not a number. Modern blue-light exposure (phones, computers, LEDs) pushes the same way: in a controlled trial, five evenings of e-reader use suppressed melatonin about 55% and delayed its onset (see the screens page below); ordinary evening screen use is a weaker version of the same effect. This explains why modern people have widely disrupted sleep rhythms, why the elderly struggle to fall asleep, and why blue-light glasses or phone night mode actually make sense.
One common misconception: 'more serotonin = better sleep' is only half right. Serotonin is melatonin's precursor, but supplementing serotonin or tryptophan alone won't significantly raise melatonin, because other regulatory steps gate the path.
机制 · MT1 / MT2 受体 · 清醒驱动怎么被摁下去
睡眠化学里的真实角色褪黑素不是直接让你睡着, 它是告诉身体现在是夜晚 的信号。睡眠本身是腺苷、睡眠压力和昼夜节律共同作用的结果。褪黑素的实际作用是调节生物钟相位 (phase shifter)、让体温降低 0.3-0.5°C, 同时减弱清醒驱动。
减弱清醒驱动 这几个字是全篇最容易被一眼扫过去的一句, 但相位调节器而不是镇静剂 这个结论的全部重量都压在它上面。拆开看是三步。
第一步 · 它落在哪里
褪黑素随血流跑遍全身, 但只在装了对应插座 的地方起作用。这个插座叫褪黑素受体, 主要两型: MT1 和 MT2, 都属于 G 蛋白偶联受体 (GPCR) 这一大类——可以想成细胞膜外侧的一个门把手, 分子一停靠上去, 细胞里面的信号链就被拨动一下。
在大脑里, 这两个插座最密集的位置恰好是两个和醒着 直接相关的地方:
SCN (视交叉上核)——总时钟本身。白天它自己在高频放电, 这串放电就是保持清醒 这条指令的源头之一视前区 (preoptic area)——下丘脑前部, 既负责发起睡眠, 也是身体体温调定点所在的地方
第二步 · 停靠之后发生了什么
褪黑素停到 suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. 神经元的 MT1 上, 细胞内那条兴奋 信使被调低, 这群神经元的放电频率跟着下降。SCN 白天那串保持清醒 的指令因此变弱——请注意方向: 不是有人按下了睡眠开关, 而是清醒这条指令自己变小了。这就是减弱清醒驱动 的字面意思, 也是它和安眠药最根本的分界: 安眠药是往刹车上加力 (增强抑制性信号), 褪黑素只是把油门松开一点。油门松了你未必立刻停车——这正是很多人抱怨吃了没感觉 的原因, 他们期待的是刹车。
停到 MT2 上的那一部分则主要负责拨时钟: 它改变 SCN 内部节律细胞的相位, 让整套时间表往前挪或往后挪。同一粒药, 服用时刻不同, 拨的方向可以相反 (见时差幕)。
第三步 · 体温为什么跟着掉
同一种受体也长在皮肤的小动脉上。褪黑素一停靠, 这些血管松开变宽, 手和脚的血流量增加, 热就顺着四肢皮肤散出去。人体核心正是靠这样把热让出去 来降温的——你在困的时候手脚发热, 不是巧合, 那恰恰是核心在往下掉。与此同时视前区那边的体温调定点也被往下调了一点, 于是身体主动配合着降。
核心体温往下掉这件事本身就是入睡的必要条件之一: 体温处在下降段 的那一小时里, 入睡的门槛最低。所以褪黑素那部分助眠 的观感, 有很大一块其实是绕道体温 实现的, 而不是直接作用在意识 上。
这三步一次解释了三件本来要死记的事
为什么它对倒时差 最有效: 时差的本质就是相位错位, 而 MT2 那条路正好是拨相位的为什么它对半夜反复醒 几乎没用: 那不是相位问题, 拨时钟拨不到那里去为什么把卧室调凉 和吃褪黑素 会出现在同一张对比表上: 它们最后落在同一个终点——把核心体温送下去。一个从外面拿走热, 一个从里面让身体自己放热
数字 · 一天里的褪黑素曲线 + 随年龄的衰减
典型节律:白天血浆水平 < 10 pg/mL晚上 9-10 点开始上升 (Dim Light Melatonin Onset, DLMO)凌晨 2-4 点是峰值 —— 这个峰的高度个体差异很大 (量级大致在每毫升几十到一百多皮克; 儿童更高, 老人显著低)早晨 6-7 点降回基线
这条曲线该怎么读
关键不是某一刻的绝对值, 而是上升发生在什么时候。DLMO 那个拐点是全身时间表的发令枪: 它往后挪几个小时, 后面所有挂在它下游的事——核心体温开始下降的时刻、深睡出现的时刻、第二天皮质醇抬头的时刻——都跟着往后挪。所以熬夜之后早上起不来 并不是意志力问题, 而是发令枪本身被推迟了, 身体只是老老实实按新时间表办事。
也正因为要读的是拐点 而不是总量, 老人峰值低这件事并不自动等于该补: 峰值低但拐点准时的人, 时间表是对的; 峰值正常却每晚被屏幕往后推两小时的年轻人, 时间表反而是乱的。
衰老和现代生活带来的变化
30 岁之后峰值随年龄下降, 到老年时明显低于年轻人 —— 但各家研究给出的降幅差得太远, 这里只给方向, 不给一个具体百分比。现代蓝光暴露 (手机、电脑、LED) 推的是同一个方向: 在一项受控试验里, 连续几个晚上的电子阅读器阅读把褪黑素压低了约 55%、并推迟了它的起点 (见下面讲屏幕那一页); 日常的晚间用屏是同一效应的较弱版本。这就解释了为什么现代人睡眠节律普遍乱、老人难入睡, 以及为什么蓝光眼镜或手机夜间模式真的有意义。
一个常见误解: 血清素更高睡得更好 只对了一半。血清素是褪黑素的前体, 但单独补血清素或色氨酸不会显著增加褪黑素, 因为合成受其它环节调控。
想看原料怎么一步步拼成褪黑素的话: 色氨酸 (Tryptophan) → 5-羟色氨酸 → 血清素 (5-HT) → N-乙酰血清素 → 褪黑素。这条线上的限速酶是 AANAT (芳基烷基胺 N-乙酰转移酶), 同样听昼夜节律的指挥——白天 suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. 摁住它, 黑暗下放开它上调, 褪黑素才能飙起来。
Blue light · screens · modern sleep debt
The real mechanism behind 'screens hurt sleep' isn't 'tired eyes' — it's chemistry.ipRGC melanopsin is a special class of retinal ganglion cells (non-visual, 1–2% of total) most sensitive to 460–480 nm blue light. They don't form images — they project directly to the suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. and accessory optic nuclei. This pathway IS the light-circadian path.
Blue-light signature of modern screens:
Phone and tablet LCD displays: blue peak ~450 nm, exactly hitting melanopsin's most sensitive bandOffice LED cool white is similar2 hours of screen exposure between 9–11 pm → SCN gets a 'still daytime' signal → AANAT suppressed → melatonin secretion delayed 1–3 hours
Chang 2015 PNAS classic RCT (N=12 healthy adults, crossover design): tablet reading vs paper reading for 5 nights × 4 hours each — tablet group melatonin secretion ↓55%, sleep onset delayed, morning alertness reduced.
Practical countermeasures (ranked by effect):
1. Stop screens 1–2 hours before bed — the strongest single intervention
2. Enable phone Night Shift / f.lux — about 50% blue-light reduction
3. Blue-blocker glasses (orange/yellow) — must actually block 480 nm, not just a 'slightly yellow' gimmick
4. Warm indoor lighting (2700–3000 K) instead of cool white
5. Morning 10 minutes of bright light — opposite of evening, anchors the SCN to 'daytime'
Dimming the screen helps a little (less intensity), but shifting color temperature is more effective (less blue band).
Children + screens + insomnia: children have higher melanopsin sensitivity and are more affected by blue light. The AAP recommends no screens under age 2, and screen-free for 1 hour before bed in school-age children. Short sleep in children and teens is consistently associated with screen use (Carter 2016 *JAMA Pediatr* meta). Note that this meta pools mainly cross-sectional studies — it can show the two travel together, not which causes which (children who already sleep badly are also more likely to reach for a screen at night). The Chang trial above is the causal-direction evidence, and what it measured was melatonin under controlled conditions, not population insomnia rates.
Chapter 2
0.3 mg vs 5-10 mg · dose myth
0.3 mg vs 5-10 mg · dose myth
Melatonin is the most ironic 'more is better' counter-example in nutritional medicine.
The physiological peak occurs between 2–4 am, but how high that peak runs varies widely between people, so using it as the denominator for 'how many times over' is not reliable. There is a firmer ruler: an oral dose of 0.1 mg is already enough to raise plasma melatonin into the normal nocturnal range (*Dollins 1994* *PNAS*). Divide the shelf doses by that number and the inversion is obvious.
0.3 mg oral is already in the physiological nocturnal range — enough to induce sleep and shift the circadian phase0.1–0.5 mg: Zhdanova 2001 *JCEM* enrolled 30 people — 15 with actigraphically confirmed reduced sleep efficiency plus 15 normal-sleeping controls — comparing 0.1 / 0.3 / 3.0 mg. 0.3 mg was most effective in the poor-sleeping group; in the normal sleepers, none of the doses changed sleep efficiency or sleep architecture — that second half is the most direct evidence that people who already sleep well get nothing from it. (Note: Zhdanova 1995 was a separate study in healthy young adults — two different papers; don't conflate them.)3 mg+: roughly 30× the dose needed to return plasma to the nocturnal range — similar effect to lower doses initially, but more side effects5–10 mg: roughly 50–100× — most RCTs show nitric oxide: A small signal molecule from the vessel lining that relaxes the vessel-wall muscle so the vessel widens. better effect, and more next-day drowsiness, headache, and endocrine disruption
Why are commercial products all 3–10 mg? Several reasons stack:
1. Historical accident — 1990s early products were dosed by 'feels like a pill', not by RCTs
2. The consumer expectation that 'more = safer'
3. Patent protection expired long ago — manufacturers differentiate by dose and packaging
4. 'Stronger action' marketing — empirically false
Why high doses may actually be worse:
Blood concentration too high → next-day residue, presenting as drowsiness, headache, a 'melatonin hangover' feelingMelatonin receptors (MT1/MT2) downregulate; chronic high-dose use may cause toleranceHormone interference — melatonin interacts with the LH/FSH/sex hormone axis; large doses may affect puberty or menstruation5+ mg side effects include vivid dreams, hypothermia, short-term mood changes
How to pick a dose:
Healthy adult under 40 + occasional jet lag: 0.3–0.5 mg, 30–60 min pre-bedChronic insomnia + 60+ elderly: start at 0.5–1 mg, observeChildren (clinical use): 0.5–2 mg under MD guidance, not dailySpecial scenarios (blind, severe circadian disorder): prescription dose + medical supervision, may be higher
4-week trial protocol:
1. 0.3 mg × 30 min pre-bed × 1 week
2. No effect, raise to 0.5 mg × 1 week
3. Still no effect, raise to 1 mg × 1 week
4. 1 mg still no effect — this is not a dose problem, it's an indication problem; see a doctor instead of escalating further
The physiological peak occurs between 2–4 am, but how high that peak runs varies widely between people, so using it as the denominator for 'how many times over' is not reliable. There is a firmer ruler: an oral dose of 0.1 mg is already enough to raise plasma melatonin into the normal nocturnal range (*Dollins 1994* *PNAS*). Divide the shelf doses by that number and the inversion is obvious.
0.3 mg oral is already in the physiological nocturnal range — enough to induce sleep and shift the circadian phase0.1–0.5 mg: Zhdanova 2001 *JCEM* enrolled 30 people — 15 with actigraphically confirmed reduced sleep efficiency plus 15 normal-sleeping controls — comparing 0.1 / 0.3 / 3.0 mg. 0.3 mg was most effective in the poor-sleeping group; in the normal sleepers, none of the doses changed sleep efficiency or sleep architecture — that second half is the most direct evidence that people who already sleep well get nothing from it. (Note: Zhdanova 1995 was a separate study in healthy young adults — two different papers; don't conflate them.)3 mg+: roughly 30× the dose needed to return plasma to the nocturnal range — similar effect to lower doses initially, but more side effects5–10 mg: roughly 50–100× — most RCTs show nitric oxide: A small signal molecule from the vessel lining that relaxes the vessel-wall muscle so the vessel widens. better effect, and more next-day drowsiness, headache, and endocrine disruption
Why are commercial products all 3–10 mg? Several reasons stack:
1. Historical accident — 1990s early products were dosed by 'feels like a pill', not by RCTs
2. The consumer expectation that 'more = safer'
3. Patent protection expired long ago — manufacturers differentiate by dose and packaging
4. 'Stronger action' marketing — empirically false
Why high doses may actually be worse:
Blood concentration too high → next-day residue, presenting as drowsiness, headache, a 'melatonin hangover' feelingMelatonin receptors (MT1/MT2) downregulate; chronic high-dose use may cause toleranceHormone interference — melatonin interacts with the LH/FSH/sex hormone axis; large doses may affect puberty or menstruation5+ mg side effects include vivid dreams, hypothermia, short-term mood changes
How to pick a dose:
Healthy adult under 40 + occasional jet lag: 0.3–0.5 mg, 30–60 min pre-bedChronic insomnia + 60+ elderly: start at 0.5–1 mg, observeChildren (clinical use): 0.5–2 mg under MD guidance, not dailySpecial scenarios (blind, severe circadian disorder): prescription dose + medical supervision, may be higher
4-week trial protocol:
1. 0.3 mg × 30 min pre-bed × 1 week
2. No effect, raise to 0.5 mg × 1 week
3. Still no effect, raise to 1 mg × 1 week
4. 1 mg still no effect — this is not a dose problem, it's an indication problem; see a doctor instead of escalating further
RCT evidence for 'right dose'
Putting the key RCT data side by side makes the picture clear.Zhdanova 2001 (*JCEM*) — age-related insomnia dose-response classic: N=30, of whom 15 had actigraphically confirmed reduced sleep efficiency and 15 slept normally as controls, comparing 0.1 / 0.3 / 3.0 mg vs placebo. 0.3 mg was most effective at restoring sleep efficiency in the poor-sleeping group; 3.0 mg produced supraphysiological peaks and still-elevated morning plasma. And in the 15 who already slept normally, no dose changed sleep efficiency or sleep architecture — give that half the same weight as the first: it is the 'if you sleep well, don't take it' evidence inside the same paper. Conclusion: 0.3 mg is the best balance of effect and side effects. (The 1995 Zhdanova paper studied healthy young adults — different study, do not conflate.)
Brzezinski 2005 *Sleep Med Rev* meta-analysis (17 RCTs, N=284):
Objective sleep onset latency shortened ~4 minutesTotal sleep time increased ~12–13 minutesSleep efficiency improved 2–3%Effect small but stable, clearly weaker than prescription hypnoticsNo dose-response relationship beyond 0.3 mg
Auger AASM 2015 clinical practice guideline: circadian sleep-wake disorders (blind, delayed phase, jet lag) — specialty first-line; general adult insomnia — not recommended as first choice (small effect); behavioral therapy (CBT-I) is first-line. Dose 0.5–3 mg, no higher. Timing depends on diagnosis: jet lag or delayed phase, 30 min pre-bed; advanced phase, morning dosing.
Herxheimer 2002 Cochrane meta (jet lag prevention): crossing 5+ time zones, ~50% effective rate; 0.5–5 mg are similarly effective — the review's only refinement is that people fell asleep faster and slept better on 5 mg than on 0.5 mg, and above 5 mg is no better. The review is stricter about timing than about dose: take it close to the target bedtime at the destination (about 10 pm to midnight); taken too early, in the daytime, it causes sleepiness and delays adaptation. Jet lag is one of melatonin's strongest indications.
Clinical evidence for melatonin in children:
Autism insomnia: 0.5–3 mg, AAP-acknowledgedADHD with comorbid insomnia: similar dosesHealthy children 'trouble falling asleep': not recommended, behavioral intervention firstUS melatonin gummy sales for kids have exploded, and poison-control-reported pediatric ingestions rose 530% between 2012 and 2021; in that same report five children required mechanical ventilation and two died (CDC 2022 MMWR). Not the melatonin itself but the combination of candy-style packaging, high doses, and easy overdose
Summary:
'More is better' is wrong; 0.3–1 mg is already the plateau'Insurance-style every night' has no evidence — possible tolerance plus endocrine interferenceFor jet lag or specific circadian disorders it works, but only with the right dose at the right time
Chapter 3
Jet lag + shift work · best indication
Jet lag + shift work · best indication
Melatonin's two most-evidenced uses are jet lag and shift-work adjustment — but most people get the 'how' wrong.
Jet lag
Core concept: melatonin is a circadian phase-shifter, not a sedative. The key question before taking it is 'do I need to advance or delay my clock?'
Advance: flying east (Beijing → NYC, need earlier sleep/wake)Delay: flying west (Beijing → SF, need later sleep/wake)Taking it at the wrong time = reversed effect
For eastward flights (3+ time zones):
After arrival, take 0.5–3 mg close to local bedtime (about 10 pm to midnight) for 2–3 nights — this is exactly where the Cochrane review pins its conclusion, and it warns in the next breath that taken too early in the day it causes sleepiness and delays adaptationPair with bright-light exposure in the morning after arrival (opposite action of melatonin)Starting to advance the clock a few days before departure is a separate pre-adaptation protocol (progressively earlier bedtimes + morning light, with melatonin taken before your own habitual bedtime). It is not covered by the Cochrane review above — don't treat the two as one conclusion
For westward flights:
Most don't need melatonin — natural delay is easierSevere cases can dose on predawn waking after arrival (delaying phase) — not routine
Herxheimer 2002 Cochrane: crossing 5+ time zones, RCT effective rate ~50% (improves 1–2 days of recovery time); <5 zones, effect is small.
Shift work
This is a complex and painful clinical scenario; melatonin can only help to a limited extent:
Post-night-shift daytime sleep: 0.5–1 mg on the morning return (about 6–7 am) helps extend daytime sleep — the same figure the atlas `shift-work-circadian` story usesNote there are two different purposes and two different doses here: that morning dose signals the start of *your* night; advancing the whole clock (rotation changes, DSPD) instead uses the lower 0.3–0.5 mg, taken about 4–6 hours before the target sleep timeWeekend reverse: transitions are chaotic, melatonin is adjunct, light management is primaryPermanent night vs rotating night: rotating is harder to adjustRecommend negotiating at least 3 weeks of stable shifts with the employer, then building a reversed circadian
General insomnia ≠ jet lag / shift work
The first-line treatment for chronic insomnia is CBT-I (cognitive behavioral therapy): multiple RCTs show its long-term efficacy exceeds any drug, and APA and AASM both recommend it as first-line. Melatonin has a small effect for general insomnia and is not first-choice. Subtyping further:
Trouble falling asleep: possibly delayed sleep phase disorder (DSPD) — melatonin useful, but needs proper diagnosisEarly waking: advanced sleep phase disorder (ASPD) — morning melatonin, a rare use caseMultiple awakenings: melatonin barely works — investigate apnea, anxiety, nocturia
So 90% of people 'just popping melatonin' are using it wrong — jet lag and specific circadian disorders are its strengths, treating general insomnia is its weakness.
Jet lag
Core concept: melatonin is a circadian phase-shifter, not a sedative. The key question before taking it is 'do I need to advance or delay my clock?'
Advance: flying east (Beijing → NYC, need earlier sleep/wake)Delay: flying west (Beijing → SF, need later sleep/wake)Taking it at the wrong time = reversed effect
For eastward flights (3+ time zones):
After arrival, take 0.5–3 mg close to local bedtime (about 10 pm to midnight) for 2–3 nights — this is exactly where the Cochrane review pins its conclusion, and it warns in the next breath that taken too early in the day it causes sleepiness and delays adaptationPair with bright-light exposure in the morning after arrival (opposite action of melatonin)Starting to advance the clock a few days before departure is a separate pre-adaptation protocol (progressively earlier bedtimes + morning light, with melatonin taken before your own habitual bedtime). It is not covered by the Cochrane review above — don't treat the two as one conclusion
For westward flights:
Most don't need melatonin — natural delay is easierSevere cases can dose on predawn waking after arrival (delaying phase) — not routine
Herxheimer 2002 Cochrane: crossing 5+ time zones, RCT effective rate ~50% (improves 1–2 days of recovery time); <5 zones, effect is small.
Shift work
This is a complex and painful clinical scenario; melatonin can only help to a limited extent:
Post-night-shift daytime sleep: 0.5–1 mg on the morning return (about 6–7 am) helps extend daytime sleep — the same figure the atlas `shift-work-circadian` story usesNote there are two different purposes and two different doses here: that morning dose signals the start of *your* night; advancing the whole clock (rotation changes, DSPD) instead uses the lower 0.3–0.5 mg, taken about 4–6 hours before the target sleep timeWeekend reverse: transitions are chaotic, melatonin is adjunct, light management is primaryPermanent night vs rotating night: rotating is harder to adjustRecommend negotiating at least 3 weeks of stable shifts with the employer, then building a reversed circadian
General insomnia ≠ jet lag / shift work
The first-line treatment for chronic insomnia is CBT-I (cognitive behavioral therapy): multiple RCTs show its long-term efficacy exceeds any drug, and APA and AASM both recommend it as first-line. Melatonin has a small effect for general insomnia and is not first-choice. Subtyping further:
Trouble falling asleep: possibly delayed sleep phase disorder (DSPD) — melatonin useful, but needs proper diagnosisEarly waking: advanced sleep phase disorder (ASPD) — morning melatonin, a rare use caseMultiple awakenings: melatonin barely works — investigate apnea, anxiety, nocturia
So 90% of people 'just popping melatonin' are using it wrong — jet lag and specific circadian disorders are its strengths, treating general insomnia is its weakness.
轮班 · 时差证据 · 一般失眠为什么不算适应症
为什么吃错时间会反效果, 而不只是没效果两个方向先记住: 往前调 (advance) 适合往东飞 (北京 → 纽约, 需要早睡早起), 往后调 (delay) 适合往西飞 (北京 → 旧金山, 需要晚睡晚起)。而往西飞的人多数不需要褪黑素 —— 自然延后本来就比提前容易; 严重的可以在抵达后凌晨醒来时服一次 (推迟相位), 但那不是常规做法。
拨相位这件事本身是有方向的: 同一粒药停在同一批受体上, 落在你生物钟的傍晚那一侧 就把时钟往前拨, 落在凌晨那一侧 就往后拨。所以在错误的时刻服用, 你不是白吃了一粒, 而是亲手把时钟往你不想去的方向推了一格——这是助眠剂根本不会有的一种失败模式, 也是褪黑素是相位调节器不是镇静剂 这句话在实操上最贵的一个推论。
Herxheimer 2002 Cochrane: 跨 5+ 时区 RCT 有效率约 50% (改善 1-2 天恢复时间), 跨 < 5 时区效应小。
轮班工作 (shift work)
这是个复杂且痛苦的临床场景, 褪黑素只能有限度地帮上忙:
夜班后白天睡眠: 早晨回家时 (约 06-07 时) 服 0.5-1 mg, 帮助延长白天睡眠 (与 atlas `shift-work-circadian` 用的是同一个口径)注意这里有两个不同用途、两个不同剂量, 别混: 上面那一粒是在你的夜开始时给 SCN 发信号; 而要把整块时钟往前拨时 (轮班换向、DSPD), 用的是更低的 0.3-0.5 mg, 且要放在目标入睡时刻前约 4-6 小时周末倒回: 转换期混乱, 褪黑素只是辅助, 主力是光照管理持续夜班 vs 旋转夜班: 旋转更难调节建议与公司协商至少 3 周稳定的班次, 再建立反向昼夜节律
为什么轮班比时差难得多
时差只需要把时钟整体挪一次, 挪完了当地的太阳会替你把它按住——光照和你的新作息是一致的, 时钟自己就锚上了。轮班没有这个便利: 你要求身体在天亮时睡、天黑时醒, 而窗外的太阳每天都在往相反方向拽。suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. 收到的最强信号始终是光, 褪黑素这一路只能在旁边小声地推。
所以轮班场景里真正的主力是遮光 (白天睡觉时把卧室做成夜晚) 和定时亮光 (上班前用强光把清醒那一侧顶起来), 褪黑素只是帮忙把那一觉的入口对准。旋转班之所以最难, 是因为它连一个稳定的目标时刻 都没有——时钟每挪一次要好几天才走完, 而班次换得比这更快。
一般失眠 ≠ 时差、轮班
慢性失眠的首选是认知行为疗法 (CBT-I): 多个 RCT 显示它在长期效果上强于任何药物, APA 和 AASM 一致推荐为一线。褪黑素对一般失眠效果小, 不是首选。再分类看一下:
入睡难: 可能是延迟期睡眠障碍 (DSPD), 褪黑素有用, 但需要明确诊断早醒: 高级期睡眠障碍 (ASPD), 褪黑素早晨服, 用法比较少见夜里多次醒: 褪黑素几乎无效, 应该排查呼吸暂停、焦虑、夜尿等
这三种分类为什么对应三种不同答案
把它们放回受体机制上就一目了然: 褪黑素能做的只有把整张时间表往前后挪 加把清醒驱动压低一点。
入睡难 (整张表偏晚) 正好是它的形状: 表挪回来, 问题就解决了早醒 (整张表偏早) 也是同一个工具, 只是方向相反, 所以要早晨服——反直觉但机制上一致夜里反复醒 完全不是表的问题: 表是准的, 是睡眠在夜里被别的东西打断了 (呼吸暂停时的憋醒、焦虑时的觉醒、膀胱)。挪表挪不掉一个每晚来敲门的东西, 所以再加量也没有用
吃了没效就加量 之所以是个陷阱, 根子就在这里: 它默认所有失眠是同一个病、只是剂量不够; 而实际上第三类根本不在这个工具的射程内。
所以随手吃片褪黑素 的人, 90% 用错了——时差和特定昼夜节律障碍是它的强项, 治疗一般失眠是它的弱项。
CBT-I · the real first-line for insomnia
Why don't most doctors reach for melatonin first? Because there's a stronger option — CBT-I (Cognitive Behavioral Therapy for Insomnia). It has six components.① Sleep restriction — the most counterintuitive but strongest:
Compress time in bed to actual sleep time, raising sleep efficiencyExample: bed at 11 pm, sleep at 1 am, wake at 6 am — 5 h sleep but 7 h in bedPrescription: time in bed = 5 hours, bed at 1 am, wake at 6 amSustain 1–2 weeks: sleep pressure accumulates, onset speeds up, efficiency risesThen relax by 15 min/week toward the optimum
② Stimulus control:
Bed is for sleep and sex only — no phone, work, or TV in bedAfter 20 minutes not asleep, get up, do something boring, return when sleepyRebuild the 'bed = sleep' conditioned reflex
③ Sleep hygiene: room 18–20°C; absolute darkness and quiet; same wake time every day even after a bad night; caffeine cutoff at 2 pm; avoid alcohol (disrupts REM); screens off 1–2 hours before bed.
④ Cognitive restructuring: the belief 'I MUST get 8 hours tonight' is the strongest catalyst of insomnia. Rewrite it as 'a bad night doesn't mean I can't function tomorrow' to break the sleep-anxiety → worse-sleep cycle.
⑤ Relaxation training: progressive muscle relaxation, mindfulness meditation, 4-7-8 breathing. The goal is not 'pre-bed relaxation' but reducing daytime hyperarousal.
⑥ Paradoxical intention (advanced): deliberately try to stay awake — releases the 'I must sleep' pressure.
RCT evidence: multiple meta-analyses show CBT-I is as effective as the strongest hypnotics (zolpidem), and superior long-term. AASM, APA, NICE, and WHO all recommend it as first-line for chronic insomnia. The regulatory status of the digital versions differs and should not be flattened into 'approved': Somryst is an FDA-cleared prescription digital therapeutic (the commercial successor to the SHUTi program used in the research — SHUTi itself was never approved), Sleepio took the UK NICE/NHS recommendation route, and CBT-I Coach is a free US Veterans Affairs app.
So 'using melatonin for chronic insomnia' before trying CBT-I means missing the stronger option. The reasonable path is CBT-I first, with short-term melatonin as adjunct if needed.
Chapter 4
Label chaos + commercial issues
Label chaos + commercial issues
Erland 2017 JCSM sampling tested 31 commercial melatonin products in Canada:
Label vs actual content differed -83% to +478%More than 71% of products fell outside ±10% of the labelled amountBetween batches of the same product, content varied by as much as 465% — even buying the same bottle every time, you don't know whether this batch matches the last oneEight products (about 26%) contained unlabelled serotonin (5-HT), at levels of 1 to 75 μg
Don't skip that μg figure — it decides how the risk should be read. The 5-HTP supplement doses that cause trouble are in the 50-300 mg range — three to four orders of magnitude away from 1-75 μg, and oral serotonin is poorly absorbed and does not cross the blood-brain barrier. So the paper's point is not that these products cause serotonin syndrome; it is that an active molecule that appears nowhere on the label is in the bottle. For someone on an SSRI or MAOI, the real problem is not knowing what you swallowed, not this particular quantity.
The children's gummies were measured by a different study: *Cohen 2023* *JAMA* analysed 25 melatonin gummy products sold in the US — one contained no melatonin at all, and the rest ranged from 74% to 347% of the labelled amount. A gummy marked 1 mg may be 0.74 mg, or it may be 3.5 mg.
FDA / NMPA / Health Canada inconsistent regulation:
USA: classified as dietary supplement, no FDA approval needed — label accuracy and purity left to manufacturer self-policingCanada: NHP (Natural Health Product) registered, stricter than USMost EU countries: prescription medication, not OTC — starts at 1 mg, requires MD prescriptionAustralia: prescriptionChina: classified as health food — labels more regulated but content consistency variesJapan: not allowed to be sold as a supplement
This means the same '3 mg melatonin capsule' bought at a US supermarket vs a European pharmacy may not be the same product at all — purity, dose, accessory ingredients, and clinical supervision all differ.
US pediatric ingestion surge: what CDC 2022 MMWR counted was poison-control-reported pediatric melatonin ingestions — 260,435 of them between 2012 and 2021, with the annual number rising 530%, driven by gummy formulation + high dose (5–10 mg) + candy-style packaging. Most children recovered, but in that same report five required mechanical ventilation and two died — those two figures are the actual weight of this scene, and 'most recover' should not cover them up.
Long-term use unknowns:
RCT data beyond 12 weeks is scarceLong-term effect on puberty and sexual development in children is unknown; concern stems from interaction with the reproductive hormone axisTolerance: some users report needing gradual escalation, RCT data are inconsistentDependence and withdrawal: psychological dependence is common, abrupt cessation may produce rebound insomnia
The boundary of 'safe':
Healthy adult, occasional use, <5 mg, <2 weeks: virtually no serious side effectsLong-term nightly 5+ mg: boundary unclear, concern exceeds evidenceChildren, pregnant women, teens: MD-guided, short-term, only when neededAnyone taking fluvoxamine: do not combine. Melatonin is cleared mainly by the liver enzyme CYP1A2, and fluvoxamine is one of its strongest inhibitors — the same tablet produces an order of magnitude more exposure, and the European prescription-melatonin label says the combination should be avoided. The same route implicates other strong CYP1A2 inhibitors (ciprofloxacin and other quinolones). This is the one interaction on the list with a hard number behind itPeople on anticoagulants, antihypertensives, immunosuppressants, or antiepileptics: caution — but to be clear, these are theoretical interactions and carry far less weight than the one above
Label vs actual content differed -83% to +478%More than 71% of products fell outside ±10% of the labelled amountBetween batches of the same product, content varied by as much as 465% — even buying the same bottle every time, you don't know whether this batch matches the last oneEight products (about 26%) contained unlabelled serotonin (5-HT), at levels of 1 to 75 μg
Don't skip that μg figure — it decides how the risk should be read. The 5-HTP supplement doses that cause trouble are in the 50-300 mg range — three to four orders of magnitude away from 1-75 μg, and oral serotonin is poorly absorbed and does not cross the blood-brain barrier. So the paper's point is not that these products cause serotonin syndrome; it is that an active molecule that appears nowhere on the label is in the bottle. For someone on an SSRI or MAOI, the real problem is not knowing what you swallowed, not this particular quantity.
The children's gummies were measured by a different study: *Cohen 2023* *JAMA* analysed 25 melatonin gummy products sold in the US — one contained no melatonin at all, and the rest ranged from 74% to 347% of the labelled amount. A gummy marked 1 mg may be 0.74 mg, or it may be 3.5 mg.
FDA / NMPA / Health Canada inconsistent regulation:
USA: classified as dietary supplement, no FDA approval needed — label accuracy and purity left to manufacturer self-policingCanada: NHP (Natural Health Product) registered, stricter than USMost EU countries: prescription medication, not OTC — starts at 1 mg, requires MD prescriptionAustralia: prescriptionChina: classified as health food — labels more regulated but content consistency variesJapan: not allowed to be sold as a supplement
This means the same '3 mg melatonin capsule' bought at a US supermarket vs a European pharmacy may not be the same product at all — purity, dose, accessory ingredients, and clinical supervision all differ.
US pediatric ingestion surge: what CDC 2022 MMWR counted was poison-control-reported pediatric melatonin ingestions — 260,435 of them between 2012 and 2021, with the annual number rising 530%, driven by gummy formulation + high dose (5–10 mg) + candy-style packaging. Most children recovered, but in that same report five required mechanical ventilation and two died — those two figures are the actual weight of this scene, and 'most recover' should not cover them up.
Long-term use unknowns:
RCT data beyond 12 weeks is scarceLong-term effect on puberty and sexual development in children is unknown; concern stems from interaction with the reproductive hormone axisTolerance: some users report needing gradual escalation, RCT data are inconsistentDependence and withdrawal: psychological dependence is common, abrupt cessation may produce rebound insomnia
The boundary of 'safe':
Healthy adult, occasional use, <5 mg, <2 weeks: virtually no serious side effectsLong-term nightly 5+ mg: boundary unclear, concern exceeds evidenceChildren, pregnant women, teens: MD-guided, short-term, only when neededAnyone taking fluvoxamine: do not combine. Melatonin is cleared mainly by the liver enzyme CYP1A2, and fluvoxamine is one of its strongest inhibitors — the same tablet produces an order of magnitude more exposure, and the European prescription-melatonin label says the combination should be avoided. The same route implicates other strong CYP1A2 inhibitors (ciprofloxacin and other quinolones). This is the one interaction on the list with a hard number behind itPeople on anticoagulants, antihypertensives, immunosuppressants, or antiepileptics: caution — but to be clear, these are theoretical interactions and carry far less weight than the one above
4 product criteria
If you must buy, there are 4 selection criteria.1. Dose ≤ 1 mg: start at 0.3–0.5 mg; avoid '5–10 mg' and 'high-strength' products. If hard to find, a pill cutter on a 1 mg tablet gives ~0.25 mg — a workable compromise.
2. Third-party certification (USP / NSF / ConsumerLab): a USP Verified mark means actual content matches the label, no heavy metals, GMP-compliant; ConsumerLab.com is a paid subscription but its testing is credible; NSF Certified for Sport is athlete-grade. Uncertified products are likely to produce Erland 2017-type problems — unreliable.
3. Clean ingredient list:
What you want: melatonin + minimal excipients (cellulose, magnesium stearate)What to avoid: 'sleep blends' mixing multiple sleep agents (cAMP / GABA / 5-HTP / passionflower / valerian, etc.) — synergy is unverified and side-effect risks stackEspecially avoid blends containing 5-HTP (5-hydroxytryptophan). The unlabelled serotonin Erland found makes the same point: the more complex the formulation, the less you can verify. (That study did not trace the serotonin back to 5-HTP raw material — these are two examples of one principle.)
4. Form and timing:
Sublingual tablets or sprays: fast absorption, short half-life, good for trouble falling asleepExtended-release: for multiple awakenings, mimics natural peak curveGummies: high-dose, easy to overdose, and *Cohen 2023* measured gummy content at 74%–347% of label (one product contained none at all) — the format itself is the least accurate tier. Keep out of children's reachTiming: 30–60 min pre-bed, not right at bedtime — it needs time to work
No brand list here — criteria instead. Nothing on this site supports 'brand X is better quality': neither Erland nor Cohen ranked by brand. More to the point, the previous screen just argued that children's melatonin products drive the surge in poisonings; naming a children's product one screen later contradicts it.
So take these three to the shelf. They still work in another country, or another year:
1. Third-party certification (USP / NSF) — meaning assayed content matches the label, no heavy metals, GMP-compliant
2. Clean ingredient list — pure melatonin plus minimal excipients, no 'sleep blend'
3. Dose per unit ≤ 1 mg
A product meeting all three is more reliable than any brand's reputation, whatever it is called. Where melatonin is prescription-only (much of the EU, the UK, Australia), going through a doctor satisfies the first two for you.
The long-term goal is NOT chronic dependence — rebuild your circadian via light exposure, fixed schedule, and CBT-I, demoting melatonin to 'emergency use'.
Chapter 5
Stronger interventions than melatonin
Stronger interventions than melatonin
Ranking sleep improvements by RCT-evidence ROI.
A-tier (strong RCT, large effect)
1. Behavioral therapy (CBT-I): far exceeds any supplement or single behavioral intervention. Sleep onset ↓50%, total sleep time ↑30–60 min, long-term (1–3 yr) durable — opposite of drugs. Digital versions (Sleepio, SHUTi) also effective.
2. Fixed wake time (circadian anchor): more important than bedtime. Same wake time every day, weekend drift <1 h — sleep pressure and circadian align, auto-tuning sleep onset.
3. Morning 10–30 min bright light exposure: outdoor > indoor (10000+ vs ~500 lux). suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. anchors 'morning', then melatonin appears automatically 14–16 hours later. Northern winter or indoor workers can use a lightbox (10000 lux × 30 min) instead.
4. Bedroom 18–20°C: core body temperature must drop to fall asleep. Too warm is the most common but overlooked cause of insomnia. A cooling mattress, fan, or AC at 18–19°C beats any supplement.
5. Caffeine cutoff at 2 pm: half-life 5–7 hours (slow CYP1A2 metabolizers 9–12 h). Coffee at 4 pm → 50% still in your body at 11 pm, disturbing deep sleep. 'I can fall asleep with coffee' doesn't equal 'good sleep quality' — deep sleep duration is often impaired.
B-tier (moderate RCT)
6. Regular aerobic exercise: 30+ min/day, not within 2–3 hours pre-bed. Multiple meta-analyses show effects comparable to mild hypnotics for chronic insomnia.
7. Limit alcohol: alcohol makes you fall asleep faster but significantly disrupts late-night REM and deep sleep. Habitual drinking accumulates chronic sleep debt.
8. End meals 3 hours pre-bed + limit fluids: digestion disturbs deep sleep, nocturia disturbs the whole night.
9. Bedroom absolutely dark and quiet: even the dim glow of an LED bulb can disturb melatonin. Earplugs, eye mask, blackout curtains.
C-tier (limited evidence + supplements)
Magnesium (citrate/glycinate, 200–400 mg): effective if deficient, small effect if repleteGlycine (3 g pre-bed): small RCTs show subjective improvementL-theanine (200 mg): 'relaxed without drowsy', good for pre-sleep anxietyValerian: mixed data, traditional in EuropeMelatonin (0.3–1 mg): first-line for jet lag and specific circadian disorders, small effect for general insomnia
Trendy/expensive supplements:
CBD: weak RCT data5-HTP: serotonin syndrome risk, do not self-supplementGABA supplements: GABA doesn't cross the blood-brain barrier, oral is essentially useless'Sleep blend' supplements: multiple ingredients, side-effect risks stack
Core wisdom: most 'insomnia' comes from modern anti-circadian habits. Fixing behavior and environment far outweighs any supplement, including melatonin itself.
A-tier (strong RCT, large effect)
1. Behavioral therapy (CBT-I): far exceeds any supplement or single behavioral intervention. Sleep onset ↓50%, total sleep time ↑30–60 min, long-term (1–3 yr) durable — opposite of drugs. Digital versions (Sleepio, SHUTi) also effective.
2. Fixed wake time (circadian anchor): more important than bedtime. Same wake time every day, weekend drift <1 h — sleep pressure and circadian align, auto-tuning sleep onset.
3. Morning 10–30 min bright light exposure: outdoor > indoor (10000+ vs ~500 lux). suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. anchors 'morning', then melatonin appears automatically 14–16 hours later. Northern winter or indoor workers can use a lightbox (10000 lux × 30 min) instead.
4. Bedroom 18–20°C: core body temperature must drop to fall asleep. Too warm is the most common but overlooked cause of insomnia. A cooling mattress, fan, or AC at 18–19°C beats any supplement.
5. Caffeine cutoff at 2 pm: half-life 5–7 hours (slow CYP1A2 metabolizers 9–12 h). Coffee at 4 pm → 50% still in your body at 11 pm, disturbing deep sleep. 'I can fall asleep with coffee' doesn't equal 'good sleep quality' — deep sleep duration is often impaired.
B-tier (moderate RCT)
6. Regular aerobic exercise: 30+ min/day, not within 2–3 hours pre-bed. Multiple meta-analyses show effects comparable to mild hypnotics for chronic insomnia.
7. Limit alcohol: alcohol makes you fall asleep faster but significantly disrupts late-night REM and deep sleep. Habitual drinking accumulates chronic sleep debt.
8. End meals 3 hours pre-bed + limit fluids: digestion disturbs deep sleep, nocturia disturbs the whole night.
9. Bedroom absolutely dark and quiet: even the dim glow of an LED bulb can disturb melatonin. Earplugs, eye mask, blackout curtains.
C-tier (limited evidence + supplements)
Magnesium (citrate/glycinate, 200–400 mg): effective if deficient, small effect if repleteGlycine (3 g pre-bed): small RCTs show subjective improvementL-theanine (200 mg): 'relaxed without drowsy', good for pre-sleep anxietyValerian: mixed data, traditional in EuropeMelatonin (0.3–1 mg): first-line for jet lag and specific circadian disorders, small effect for general insomnia
Trendy/expensive supplements:
CBD: weak RCT data5-HTP: serotonin syndrome risk, do not self-supplementGABA supplements: GABA doesn't cross the blood-brain barrier, oral is essentially useless'Sleep blend' supplements: multiple ingredients, side-effect risks stack
Core wisdom: most 'insomnia' comes from modern anti-circadian habits. Fixing behavior and environment far outweighs any supplement, including melatonin itself.
A 级 · 强 RCT、效应最大的五件事
A 级 (强 RCT, 效应大)1. 行为治疗 (CBT-I): 远超任何补剂和单一行为干预。入睡延迟 ↓ 50%, 总睡眠时间 ↑ 30-60 分钟, 长期效果 (1-3 年) 持久——这一点与药物正好相反。数字化版本 (Sleepio / SHUTi) 也有效。
2. 固定起床时间 (锚定昼夜节律): 比上床时间更重要。每天同一时间起床, 周末偏差 < 1 小时, 睡眠压力和昼夜节律就会重合, 自动调入睡时间。
3. 晨起 10-30 分钟亮光暴露: 户外 > 室内 (10000+ lux vs 约 500)。suprachiasmatic nucleus: The brain's master clock — set by light, it runs the body's day–night rhythm. 锚定早晨, 14-16 小时后自动出现褪黑素分泌。北方冬季或室内工作者可以用 lightbox (10000 lux × 30 min) 替代。
4. 卧室温度 18-20°C: 身体核心温度需要降低才能入睡。太热是最常见但被忽视的失眠原因。凉爽床垫、风扇、空调 18-19°C 比任何补剂都强。
5. 咖啡因截止下午 2 点: 半衰期 5-7 小时 (CYP1A2 慢代谢者 9-12 小时), 下午 4 点喝咖啡, 晚 11 点仍有 50% 在体内干扰深睡。我喝咖啡也能睡着 不等于睡眠质量好, 深睡时长往往受损。
为什么恰好是这五件
把它们和上一幕的受体机制对起来看, 会发现它们各自堵在链条的不同一环, 而且几乎没有重叠:
固定起床 + 晨起亮光 动的是链条最上游: ipRGC 把天亮了 报给 SCN, SCN 由此校准当天的发令时刻。晨光和晚上的蓝光是同一条通路、相反方向——所以它们不是多晒点太阳有益健康 那种泛泛建议, 而是直接在给时钟对表卧室调凉 动的是最下游的那一环: 褪黑素让皮肤血管松开、把热散掉, 核心体温才降得下来; 房间太热的时候, 热散不出去, 这一步被卡住——此时你补多少褪黑素, 信号都传到了, 只是执行不了咖啡因截止时间 动的是另一条完全独立的线: 腺苷。腺苷是睡眠压力 那一路, 咖啡因去占住腺苷的位置, 于是压力攒够了身体却收不到通知。这条线和褪黑素并联, 谁都替不了谁CBT-I 动的是清醒驱动 本身: 白天的过度警觉、对睡不着的焦虑, 都是在往清醒那一侧加码。褪黑素只能把清醒驱动往下压一点, 而 CBT-I 是把加码的手撤掉
所以这五件事不是五个可选项挑一个, 而是链条上五个不同的堵点。你堵在哪一环, 就该修哪一环——这也是为什么单吃一粒药很难覆盖: 它只对得上其中一环。
B 级 + C 级 · 中等证据与补剂货架
B 级 (中等 RCT)6. 规律有氧运动: 每天 30+ 分钟, 不要太靠近睡前 2-3 小时。多个 meta-analysis 显示其改善慢性失眠的效果与温和安眠药相当。
7. 限酒: 酒精让你更快入睡, 但显著扰乱后半夜的 REM 和深睡。习惯性饮酒等于积累慢性睡眠债。
8. 餐前结束 3 小时前 + 不大量喝水: 消化干扰深睡, 夜尿干扰整夜。
9. 卧室绝对黑暗加安静: 哪怕 LED 灯泡的微光也能干扰褪黑素。建议用耳塞、眼罩、遮光窗帘。
C 级 (有限证据 + 补剂)
镁 (citrate / glycinate, 200-400 mg): 缺镁者有效, 充足者效果小甘氨酸 (3 g 睡前): 小 RCT 显示主观睡眠改善L-theanine (200 mg): 放松不嗜睡, 适合睡前焦虑缬草 (Valerian): 数据混杂, 欧洲传统使用褪黑素 (0.3-1 mg): 时差和特定昼夜节律障碍一线, 一般失眠效果小
贵、网红补剂:
CBD: RCT 数据弱5-HTP: 有 5-HT 综合征风险, 不推荐自服GABA 补剂: GABA 不过血脑屏障, 口服几乎无用睡眠组合 补剂: 多种成分混合, 副作用风险叠加
为什么口服 GABA 几乎没用——这一条值得单独看
它是整张货架上最容易被识破、也最常被卖的一个。GABA 确实是大脑里主要的抑制性 信号, 说它让神经安静下来 没有错。问题出在它到不了现场: 脑血管的内皮细胞之间接得极密, 中间还围了一层胶质细胞的脚, 这道墙 (血脑屏障) 只放行两类东西——脂溶性足够高、能直接溶穿细胞膜的; 或者有专门运输蛋白来接的。GABA 两条都不占: 它带电、水溶, 而脑侧并没有一台把它从血里往脑里搬的运输机 (方向恰好相反, 脑里的 GABA 是就地合成的)。
所以你吞下去的 GABA 停在血里, 最后被代谢掉, 脑子那侧的浓度基本没动。少数人报告的放松感 更可能来自肠道神经末梢的外周作用和期待效应, 而不是补进了大脑。
这条推理值得记住的原因不是 GABA 本身, 而是它给了你一把可复用的尺子: 看到任何直接补某种神经递质 的宣传, 先问一句——这东西过得了血脑屏障吗? 同一把尺子也能量血清素 (见上一幕的合成链: 补前体不等于补成品)。
核心智慧: 大多数失眠 问题来自现代生活的反生物钟习惯。修复行为和环境的效果, 远远大于任何补剂, 包括褪黑素本身。
'3-day sleep reset' protocol
If your sleep is bad, here is a no-supplement 3-day reset.Day 1 · Unplug + anchor wake time
After 8 pm, no overhead lights — use table lamp with warm-color bulbs9 pm TV off, phone away10 pm set bedroom to 18–19°CDon't check the time; turn the alarm clock aroundTomorrow's wake time = your weekday wake time, even on a weekend
Day 2 · Morning light + no makeup sleep
Within 30 minutes of waking, get 10–20 minutes of outdoor light (overcast still >1000 lux outdoors)No makeup sleep, no matter how poorly you slept the night beforeNaps <30 minutes, never after 2 pmNo caffeine after 2 pmDinner 6–7 pm, then water onlyScreens off by 9 pmBed at 10:30
Day 3 · Consolidate + assess
Repeat Day 2's protocolLog three things each night: time you fell asleep, number of awakenings, how you felt in the morningImprovement ≥30%: you found the right lever — run the pattern another 2 weeks. No 'X% of insomniacs improve' figure here: nothing on this site supports oneNo improvement: workup OSA (snoring, daytime fatigue), or see a doctor for CBT-I or psychiatric evaluation
Key principles:
No reliance on supplements or drugsBehavior first, not 'pop a pill at night'Morning light is the single strongest interventionFixed wake time matters more than fixed bedtime
Why is this stronger than 5 mg melatonin? Because melatonin tunes only one variable (the signal), while insomnia is usually a multi-variable problem (circadian + hygiene + environment + psychology). One pill can't fix 6 variables.
Chapter 6
Summary · should I take it
Summary · should I take it
Practical decision checklist for melatonin.
Suitable (consider 0.3–1 mg, 30–60 min pre-bed):
Jet lag across 5+ time zones: strongest indicationLong-term shift work: paired with light managementDiagnosed delayed sleep phase disorder (DSPD): MD-guidedBlind non-24-hour rhythm disorder: strong evidence — but keep two things apart. What the FDA approved is tasimelteon (brand name Hetlioz), an MT1/MT2 receptor agonist; it is not melatonin. Melatonin itself is off-label here and needs specialist supervision. Don't read this line as 'melatonin is FDA-approved' — the point this story keeps making is that in the US the FDA classifies melatonin as a dietary supplement and neither approves nor regulates it65+ elderly + low endogenous melatonin + insomnia: a 1 mg trial is worthwhileAutism / ADHD children with comorbid insomnia: MD-guided, 0.5–3 mg
Not suitable:
'Insurance-style every night' healthy adults: no evidence, possible tolerance or dependenceChronic insomnia without CBT-I trial: missing the stronger optionHealthy children with 'trouble falling asleep': behavioral intervention, not gummiesPregnancy and lactation: insufficient safety data, use cautionAnyone on fluvoxamine or another strong CYP1A2 inhibitor (ciprofloxacin and other quinolones): do not combine — melatonin is cleared mainly by CYP1A2, and the combination raises exposure by an order of magnitude; the European prescription-melatonin label says the combination should be avoidedPatients on anticoagulants, antihypertensives, antiepileptics, or immunosuppressants: interaction risk (these are theoretical, and softer than the line above)Mixed insomnia (onset + early + multiple wakings): melatonin barely works — find the causeDepression or anxiety diagnosis with insomnia as a symptom: treat the primary condition, not melatonin
Dose principles:
Start 0.3–0.5 mg (rational low dose)Up to 1 mg if neededRarely exceed 3 mg (loses dose-response)Absolutely avoid taking 5–10 mg 'strong' products nightly, long-term (no benefit, more side effects). The one exception is jet lag: in the Cochrane data 0.5–5 mg are similarly effective, so a short trip on 5 mg is not out of bounds — but that is a few nights, not every night
Timing:
Most indications: 30–60 minutes before bedJet lag: close to local bedtime after arrival (about 10 pm to midnight) — do not move it up to duskDSPD / rotation changes (advancing the clock): about 4–6 hours before the target sleep time — the same figure the atlas `insomnia` and `shift-work-circadian` stories use; the exact hour is best set by a clinician from your DLMODaytime sleep after a night shift: 0.5–1 mg on the way home (about 6–7 am)
Relationship to other interventions: pre-bed blue-light exposure causes more damage than melatonin can repair; stop screens first. Morning light + fixed wake time > melatonin. Bedroom environment and bedding > melatonin. The first-line treatment for chronic insomnia is always a 6–8 week course of CBT-I.
Handling long-term use:
2–4 weeks: reassess whether you still need it3 months: re-evaluate, most should shift to behavioral approachesBeyond 6 months continuous nightly use: not recommended, RCT data is void
Final line: melatonin is NOT a 'sleep hormone supplement' — it's a circadian phase shifter. Used in the right scenario at the right dose, it works; treated as a universal sleep aid, it disappoints.
Suitable (consider 0.3–1 mg, 30–60 min pre-bed):
Jet lag across 5+ time zones: strongest indicationLong-term shift work: paired with light managementDiagnosed delayed sleep phase disorder (DSPD): MD-guidedBlind non-24-hour rhythm disorder: strong evidence — but keep two things apart. What the FDA approved is tasimelteon (brand name Hetlioz), an MT1/MT2 receptor agonist; it is not melatonin. Melatonin itself is off-label here and needs specialist supervision. Don't read this line as 'melatonin is FDA-approved' — the point this story keeps making is that in the US the FDA classifies melatonin as a dietary supplement and neither approves nor regulates it65+ elderly + low endogenous melatonin + insomnia: a 1 mg trial is worthwhileAutism / ADHD children with comorbid insomnia: MD-guided, 0.5–3 mg
Not suitable:
'Insurance-style every night' healthy adults: no evidence, possible tolerance or dependenceChronic insomnia without CBT-I trial: missing the stronger optionHealthy children with 'trouble falling asleep': behavioral intervention, not gummiesPregnancy and lactation: insufficient safety data, use cautionAnyone on fluvoxamine or another strong CYP1A2 inhibitor (ciprofloxacin and other quinolones): do not combine — melatonin is cleared mainly by CYP1A2, and the combination raises exposure by an order of magnitude; the European prescription-melatonin label says the combination should be avoidedPatients on anticoagulants, antihypertensives, antiepileptics, or immunosuppressants: interaction risk (these are theoretical, and softer than the line above)Mixed insomnia (onset + early + multiple wakings): melatonin barely works — find the causeDepression or anxiety diagnosis with insomnia as a symptom: treat the primary condition, not melatonin
Dose principles:
Start 0.3–0.5 mg (rational low dose)Up to 1 mg if neededRarely exceed 3 mg (loses dose-response)Absolutely avoid taking 5–10 mg 'strong' products nightly, long-term (no benefit, more side effects). The one exception is jet lag: in the Cochrane data 0.5–5 mg are similarly effective, so a short trip on 5 mg is not out of bounds — but that is a few nights, not every night
Timing:
Most indications: 30–60 minutes before bedJet lag: close to local bedtime after arrival (about 10 pm to midnight) — do not move it up to duskDSPD / rotation changes (advancing the clock): about 4–6 hours before the target sleep time — the same figure the atlas `insomnia` and `shift-work-circadian` stories use; the exact hour is best set by a clinician from your DLMODaytime sleep after a night shift: 0.5–1 mg on the way home (about 6–7 am)
Relationship to other interventions: pre-bed blue-light exposure causes more damage than melatonin can repair; stop screens first. Morning light + fixed wake time > melatonin. Bedroom environment and bedding > melatonin. The first-line treatment for chronic insomnia is always a 6–8 week course of CBT-I.
Handling long-term use:
2–4 weeks: reassess whether you still need it3 months: re-evaluate, most should shift to behavioral approachesBeyond 6 months continuous nightly use: not recommended, RCT data is void
Final line: melatonin is NOT a 'sleep hormone supplement' — it's a circadian phase shifter. Used in the right scenario at the right dose, it works; treated as a universal sleep aid, it disappoints.
Real effect vs placebo recognition
How to tell whether melatonin really works for you.Objective marker: sleep onset latency <30 min (vs your baseline)Subjective feel: morning alert, not drowsySustained for over a week — not a one-night 'good mood made me sleep' effect
Signature of placebo effect: the expectation 'tonight I will sleep' reduces anxiety, which naturally lets you fall asleep; ANY pill produces about 30% subjective improvement; but this effect isn't durable and fades once the novelty wears off.
Signs that it's really working:
Cross-time-zone use (the mechanism makes sense)Specific circadian disorder + correct timing65+ with naturally low endogenous melatonin≥4 weeks sustained improvementSymptoms return after stopping (not 'withdrawal', but the underlying indication persisting)
Self-assessment questions:
1. Trouble falling asleep, or multiple awakenings? Trouble falling asleep — melatonin may help; multiple awakenings — barely
2. Late-sleep-late-rise, or fatigued all day? Late-sleep-late-rise (DSPD) → melatonin + morning light; all-day fatigue → workup OSA, hypothyroid, depression, iron, B12
3. Sleep only 5–6 hours and not tired? Some genotypes need only 6 hours (short sleep syndrome, ~1–3% of people) — not a disease, no treatment needed
Clear signals to stop melatonin:
4 weeks of use, no subjective changeEscalated to 3 mg+ and still no effectNext-day drowsiness, headache, or low mood appearsVivid nightmares clearly increaseTachycardia or palpitationsConflict with required medications (warfarin, etc.)
When you find it 'doesn't work', don't keep escalating — return to behavioral basics and investigate the real cause of your sleep problem. A single pill can't fix an environmental, circadian, psychological, and physiological multi-factor disorder.
References · 11
- Brzezinski, A., et al. (2005). Effects of exogenous melatonin on sleep: a meta-analysis. Sleep Medicine Reviews, 9(1), 41–50. 10.1016/j.smrv.2004.06.004
- Cajochen, C., Kräuchi, K., & Wirz-Justice, A. (2003). Role of melatonin in the regulation of human circadian rhythms and sleep. Journal of Neuroendocrinology, 15(4), 432-437. Review of endogenous melatonin as the internal signal of biological night: secretion is driven by the suprachiasmatic nucleus and suppressed by light, and the hormone feeds back on the SCN through MT1 and MT2 receptors, with sleep-promoting effects that track its thermoregulatory action rather than a hypnotic one. 10.1046/j.1365-2826.2003.00989.x
- Chang, A. M., Aeschbach, D., Duffy, J. F., & Czeisler, C. A. (2015). Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. Proceedings of the National Academy of Sciences, 112(4), 1232-1237. In a controlled in-laboratory crossover, reading from a light-emitting eReader for four hours before bed across five consecutive evenings suppressed evening melatonin, delayed circadian timing and sleep onset, reduced REM sleep, and left participants less alert the next morning, compared with reading a printed book. 10.1073/pnas.1418490112
- Auger, R. R., et al. (2015). Clinical practice guideline for the treatment of intrinsic circadian rhythm sleep-wake disorders. Journal of Clinical Sleep Medicine, 11(10), 1199–1236. 10.5664/jcsm.5100
- Zhdanova, I. V., Wurtman, R. J., Regan, M. M., Taylor, J. A., Shi, J. P., & Leclair, O. U. (2001). Melatonin treatment for age-related insomnia. The Journal of Clinical Endocrinology & Metabolism, 86(10), 4727–4730. 10.1210/jcem.86.10.7901
- Dollins, A. B., Zhdanova, I. V., Wurtman, R. J., Lynch, H. J., & Deng, M. H. (1994). Effect of inducing nocturnal serum melatonin concentrations in daytime on sleep, mood, body temperature, and performance. Proceedings of the National Academy of Sciences, 91(5), 1824-1828. Twenty healthy men received 0.1, 0.3, 1.0 or 10 mg of oral melatonin at noon. Doses from 0.1 mg upward raised daytime serum melatonin into the normal nocturnal range and shortened sleep latency; the authors found no dose-response advantage to the larger doses. 10.1073/pnas.91.5.1824
- Herxheimer, A., & Petrie, K. J. (2002). Melatonin for the prevention and treatment of jet lag. Cochrane Database of Systematic Reviews, (2), CD001520. 10.1002/14651858.CD001520
- Erland, L. A. E., & Saxena, P. K. (2017). Melatonin natural health products and supplements: presence of serotonin and significant variability of melatonin content. Journal of Clinical Sleep Medicine, 13(2), 275–281. 10.5664/jcsm.6462
- Lelak, K., Vohra, V., Neuman, M. I., Toce, M. S., & Sethuraman, U. (2022). Pediatric melatonin ingestions — United States, 2012–2021. MMWR Morbidity and Mortality Weekly Report, 71(22), 725–729. 10.15585/mmwr.mm7122a1
- Cohen, P. A., Avula, B., Wang, Y. H., Katragunta, K., & Khan, I. (2023). Quantity of melatonin and CBD in melatonin gummies sold in the US. JAMA, 329(16), 1401-1402. Chemical analysis of 25 melatonin gummy products: one contained no melatonin at all, and the quantity of melatonin in the rest ranged from 74% to 347% of the labelled amount. 10.1001/jama.2023.2296
- Härtter, S., Grözinger, M., Weigmann, H., Röschke, J., & Hiemke, C. (2000). Increased bioavailability of oral melatonin after fluvoxamine coadministration. Clinical Pharmacology and Therapeutics, 67(1), 1-6. Coadministration of fluvoxamine, a strong CYP1A2 inhibitor, raised oral melatonin exposure by roughly an order of magnitude. Melatonin is cleared mainly by CYP1A2, so the interaction extends in principle to other strong inhibitors of that enzyme. 10.1067/mcp.2000.104071