Sleep supplements, ranked by how much evidence exists

Melatonin, and it is not close. Twelve separate reviews of the trials agree that it shortens the time it takes to fall asleep and adds to total sleep.¹ They also disagree about whether that change is big enough to feel.¹ Everything else on the shelf has a fraction of that evidence, and some of it has none.

How this ranking works

This ranks by how much trial evidence sits behind each product, and by how far that evidence can be trusted. It does not rank by how well anything works. The two orders are not the same.

A product with dozens of trials and a small, dependable effect ranks above one with a couple of glowing trials, because a couple of trials cannot separate a real effect from luck. A product with no combined evidence at all sits at the bottom, whatever its label promises.

One more rule holds all of it together. Sleep answers to expectation, so the only comparison worth anything is against a dummy pill.

1. Melatonin: the most evidence, the smallest claims

Melatonin has been studied enough that people now review its reviews. One such count found twelve reviews of melatonin itself, of quality ranging from moderate down to critically low.¹ They agree it helps people fall asleep sooner and sleep longer. They do not agree that the difference is one a person would notice.¹

Put a number on it. Across 23 trials, melatonin improved scores on the Pittsburgh sleep questionnaire, where lower is better, by 1.24 points.² That average hides a lot. The gain reached 2.20 points in people with respiratory disease and 2.74 in people with metabolic disorders, fell to 0.67 in those with a primary sleep disorder, and disappeared in mental and neurodegenerative illness.²

Set against prescription drugs it shrinks again. The largest comparison of insomnia treatments in adults, built on 170 trials and 47,950 participants, found four prescription sleeping drugs more effective than melatonin and concluded that melatonin brought no material benefit.³

In dementia the line is flatter still. Doses up to 10 mg had little or no effect on any major sleep measure over eight to 10 weeks.⁴ Night-time sleep rose by 10.68 minutes on average, and the range around that runs from a quarter of an hour lost to a little over half an hour gained.⁴

So the best-evidenced product on the shelf gets people to sleep a little sooner, adds a little to the total, does nothing material for diagnosed insomnia against the field, and nothing at all in dementia. That is what a large pile of evidence looks like when the effect is small.

2. Chinese herbal formulas: many trials, hard to trust

By raw count, traditional Chinese formulas have more trials behind them than anything except melatonin. One comparison put 186 trials and 13 herbal preparations on the same scale, judged those studies at low to moderate risk of bias, and ranked three formulas at the top for sleep quality.⁵ It also reported that most of the formulas beat conventional drugs on effectiveness and on safety at once.⁵

That last finding is why the count needs handling with care. A body of trials in which herbal remedies win on benefit and on side effects simultaneously is either a major discovery or a literature with a problem. A ranking cannot tell those two apart. The volume is real. What it demonstrates is not.

3. Valerian: studied for decades, still inconsistent

Valerian is the most-studied Western herb for sleep. One review gathered 60 studies covering 6,894 people, and could combine only ten of them for sleep quality and eight for anxiety.⁶ The results pointed different ways in different trials.

The explanation offered was the product itself. Valerian extracts vary, and preparations made from whole root or rhizome gave more reliable effects than the rest.⁶ No severe side effects turned up across ages 7 to 80.⁶

That is the valerian literature in one line: safe, inconsistent, and dependent on which bottle you happened to buy. The remedy proposed was better quality control of the extracts, not more trials.⁶

4. Ashwagandha: promising and small

Ashwagandha is the newest arrival with combined trial evidence behind it. Five trials, holding 254 people between them, showed it cut anxiety scores by 5.96 points.⁷ It also improved how quickly people fell asleep, how long they slept, how much of their time in bed was spent asleep, and their overall sleep-quality score.⁷ It did not reduce time spent awake during the night.⁷

Those five trials barely agreed with each other on the size of any of it.⁷ Two hundred and fifty-four people is a promising start and nothing more.

5. Everything else

Most of the shelf sits here. Magnesium, glycine, L-theanine, GABA, apigenin, tart cherry and CBD have small trials, inconsistent designs, or none at all, and none of them appears in the combined evidence behind this ranking. The absence is their position in it.

That is not the same as evidence that they fail. Untested and ineffective are different claims. Nobody has run the study that would tell you either way, a confident label is filling a gap the research has not, and neither claim is a reason to buy.

Falling asleep versus staying asleep

Almost everything above is about getting to sleep. Staying asleep is a separate measurement: how much of the night you spend awake after first dropping off, and how often you surface.

That is where the shelf goes quiet. In dementia, melatonin showed no effect on time awake in the night, on the number of awakenings, or on the share of time in bed spent asleep.⁴ Ashwagandha improved falling asleep and total sleep, but not time awake in the night.⁷ The melatonin reviews speak about falling asleep and total sleep and stay silent on holding sleep together.¹

No supplement here has combined evidence for keeping you asleep.

What remains unknown

Whether melatonin's effect is large enough to matter is still open, and twelve reviews could not settle it between them.¹

For the herbal formulas, the question is whether those trials would come out the same way run somewhere else, and nothing in this evidence answers it.⁵

Valerian's results will stay inconsistent while its extracts do.⁶ Ashwagandha needs trials several times the size of the ones it has.⁷

And for the rest of the shelf, the honest position is not that it failed. It is that it has not been tested well enough to rank.