The bottle in the bathroom cabinet says sleep. Three milligrams, ten milligrams, gummy or tablet, bought without a prescription in much of the world and swallowed by millions of people who want the night to go faster. For families living with dementia, it is often the first thing tried when a parent starts wandering the hallway at two in the morning, and often the first thing quietly abandoned when nothing much changes. That framing, melatonin as a mild sedative, is precisely what a new review in the journal Aging and Disease sets out to unsettle.
The review, titled "Melatonin in Circadian Rhythm and Brain Health: Neuroprotective Effects and Therapeutic Potential" and carrying a DOI stamped 2026.0119, is a synthesis rather than a new experiment: no fresh participants, no new trial, no result of its own. Its authors gather work on melatonin across circadian rhythm regulation, Alzheimer's disease, Parkinson's disease, cerebral ischemia and reperfusion, depression, insomnia, brain aging and dietary sources, and they describe the hormone as a pineal product that does considerably more than make people drowsy: antioxidant, anti-inflammatory, immune-modulating. The case being made is that melatonin is a timing signal the aging brain gradually loses, and that restoring the signal is a different therapeutic proposition from blunting wakefulness. The review is not shy about the state of the file. Much of what it summarises, the authors write, comes from small clinical trials or animal models that still require validation, and their conclusion is hedged in a phrase families should hold onto: "evidence strength varies across conditions".
This is where our section earns its name. If melatonin works as a clock drug, then the measurements that matter are clock measurements: when the body's own melatonin rises relative to when the dose is given, whether the rest and activity cycle flattens or sharpens, whether the day has a shape. Those are not the numbers that dominate the practical literature in dementia. What dominates is total sleep time and night-time awakenings, endpoints that answer a caregiver's most urgent question but not the mechanistic one the review is posing. A synthesis can argue that a pathway is real. It cannot, on its own, tell you that giving the pathway a nudge changes how a person lives.
The dementia-specific record remains thin, and the most rigorous account of it is a decade old. The Cochrane review of drug treatments for sleep disturbance in dementia, led by Jenny McCleery and last searched to March 2016, identified six randomised placebo-controlled trials with 326 participants in total, covering three drugs: melatonin in four trials, trazodone in one, ramelteon in one. Almost all of the people in the melatonin and trazodone trials had moderate to severe dementia. The reviewers judged that overall "the evidence was of low quality", their standard phrasing for a body of work that further research is very likely to change. Six trials and 326 people is not a foundation you would want to build a treatment decision on, and Cochrane said as much.
So two documents sit side by side, and they do not contradict each other. One says the biology is interesting, plural and worth pursuing. The other says the human trials in dementia were few, small and weakly powered when they were last counted. Neither says melatonin protects the brain. A separate review published in Frontiers in Neuroscience in March 2026 makes the general point that circadian rhythms decay with age and that the decay travels with sleep disruption and cognitive decline, but travelling together is not the same as one causing the other, and no observational pattern in this field has yet been converted into a demonstrated benefit.
What this means at the kitchen table is narrower than the headline instinct suggests, and more useful. Melatonin is not a neutral substance simply because it is sold over a counter: dose, formulation and, if the circadian argument is right, the hour it is taken, all plausibly matter, and none of that is settled. Timing a dose to a person's own rhythm is a clinical judgement involving their other medicines and their diagnosis, not a supermarket decision. If a family is already using it, or thinking about it, the conversation belongs with the prescribing clinician, who can also say what would count as it working and how long to give it before deciding. The review's real contribution is to put a research question on the table: measure the clock, not just the sleep.