A wet city street at night seen from above, sodium lamps smeared across the asphalt

It mostly covers the street, and that is not a small thing. None of the systematic reviews in this area include a trial of rain sound as its own condition, so claims about rain specifically are borrowed from research on generated broadband noise, where the evidence was graded very low.

That sentence is going to be less satisfying than what the search results promise, so it is worth being precise about where the evidence is strong and where it is absent. The better-supported part is that people living with unpredictable nighttime noise report disturbed sleep in proportion to how loud it is. The absent part is the entire category of studies about rain.

What research exists on rain sounds specifically?

As far as I can find, none, and this is the single most useful thing on this page.

Search the sleep literature for rain and you will find rain as a weather exposure, rain as an environmental variable, and rain as a stimulus in a handful of small studies where it appears alongside four other sounds. What you will not find is a trial that took a population, played them rain, played a control group something else, and measured sleep. The category of research that gets cited whenever rain comes up is research on generated noise: white, pink and brown, produced by a machine or an algorithm.

The Sleep Foundation’s own page on white noise is unusually direct about the taxonomy. Brown noise is the color people compare to rainfall or a shower running, and green noise is the one compared to a stream or ocean waves. Both are described as having limited research behind them, and the page says plainly that researchers do not yet know how white noise improves sleep for some people, offering masking as one theory rather than an established mechanism.

So when a page tells you rain sounds are proven to help you sleep, the proof it is gesturing at is about a different sound, produced a different way, and it is weaker than the sentence implies. If you want the full picture of the colors, the noise colors overview and the brown noise vs white noise comparison go through them properly.

How weak is the evidence for noise as a sleep aid?

Graded very low, by the review that looked hardest.

The reference point is the 2021 systematic review by Riedy, Smith, Rocha and Basner at the University of Pennsylvania, which is covered against the music alternative in the post on the best sounds for sleep. The number to carry over here is the grade: across 38 articles on continuous noise, applying the GRADE criteria, the authors rated the quality of evidence for continuous noise improving sleep as very low, and said in the same breath that this contradicts how widely it is used.

The disagreement between studies was not marginal either. It ran to the extremes, with some finding that continuous noise improved sleep and others that it disrupted it.

Very low certainty does not mean the thing does not work. It means the studies were too few, too small or too inconsistent to tell you. That is a different verdict from a negative one, and it is the honest place to stand.

Then why does it work for me?

Probably because of what it is covering, and that part has real evidence behind it.

In 2022, Smith, Cordoza and Basner updated the World Health Organization’s systematic review on environmental noise and sleep. Adding 11 new studies covering 109,070 responses to the 25 original studies covering 64,090, they found a dose-response relationship in self-reported sleep disturbance. Where the survey question named noise as the source of the disturbance, the evidence was moderate quality and the odds of being highly sleep-disturbed rose by a factor of about 2.18 for aircraft and 2.52 for road traffic per 10 decibel increase in nighttime noise. Where the question did not mention noise, the odds ratios fell to 1.52 and 1.14 and the evidence was graded low to very low, which is a caveat worth carrying.

That review is about the problem, not the solution, and it is built entirely on what people reported rather than on what a sleep lab recorded. It establishes that variable outside noise tracks with worse self-reported sleep, which is the premise masking rests on. It does not test whether covering that noise fixes it.

The closest anyone has come to testing the fix directly is small. Ebben, Yan and Krieger studied ten sleep-clinic patients in New York City who reported noise-disturbed sleep, running a week of baseline, a week with a white noise machine, and a week of washout. Wake after sleep onset improved significantly by actigraphy. Ten people, one city, one device, no rain condition. It is a real result and it is nowhere near enough to generalize from, which is why the reviews above still say what they say.

Put together, the defensible version is this: if something outside is waking you, a steady sound that covers it is doing something the research can account for. If nothing outside is waking you, the case for adding a sound is much thinner than the internet suggests. The post on the best sounds for sleep works through the alternatives on the same basis.

What about rain as something you simply like?

That is a legitimate reason, and it does not need a study.

Preference is not evidence, but it is not nothing either. Adherence was one of the sources of variation the Penn review noted across studies, which is a formal way of saying people stop using sleep aids they find irritating. If rain is the sound you will actually leave on, that matters more than which color noise scored better in a trial of forty people.

There is also a plainer point about what rain is. Recorded rain is dense, broad and slow-moving, which is exactly the acoustic profile that covers speech and traffic well. It is a good masker on physical grounds, whatever the sleep literature has failed to measure about it.

How loud, and should it run all night?

Low, and probably not all night.

The National Institute on Deafness and Other Communication Disorders gives the threshold: sounds at or below 70 A-weighted decibels are unlikely to cause hearing loss even after long exposure, while long or repeated exposure at or above 85 dBA can cause it. The Sleep Foundation adds the practical failure mode, noting research that found nearly 65 percent of infant white noise machines and apps were capable of playing louder than recommended when the sleeper was very close to the device. The devices are built to go louder than anyone needs.

Set it just above the intrusion you are covering and no higher. If the street is quiet tonight, turn it down or turn it off, because the volume that masks a garbage truck is louder than the volume you need when there is no truck.

This is the point to say what this app is. Pocket Lofi plays original lofi and chillhop music in five stations, Focus, Rain, Sleep, Chill and Groove. It does not generate white, pink or brown noise. The Rain station is recorded music with rain in it rather than a rain generator, the sleep timer fades out instead of cutting off, and background play means you can turn it down from the pillow. If what you want is an eight-hour rain generator, that is a different product and you should get that one.

None of this makes rain a worse choice than it was before you read it. It makes the claim smaller and more accurate: a dense, steady sound covers an unpredictable one, which the evidence supports, and everything beyond that is preference. If you want the longer version of where the music side of this comes from, the lofi guide picks up there.

Sources