You can get better at remembering your dreams. Almost all of the work happens in the first couple of minutes after you wake up, and that window is shorter than most advice admits.

Here is what the research supports, what it does not, and one very popular tip that turns out to rest on nothing.

How Fast the Window Closes

The clearest measurement is old and still stands. In 1957, William Dement and Nathaniel Kleitman woke sleepers in a lab and asked what had been going through their minds. Waking someone during REM sleep, the stage with the most vivid dreaming, produced a dream report 152 times out of 191. Waking them from non-REM sleep produced one 11 times out of 160.

A smaller number in the same study matters more to you. When they woke people from non-REM sleep more than eight minutes after a REM period had finished, they got a report in 6 tries out of 132, as the study is usually summarised. Eight minutes of the brain doing something else was enough to empty the room. So when a dream slips away while you are hunting for your glasses, that is a clock, not a weak memory.

Why the Memory Is Weak in the First Place

Part of the answer is a chemical called norepinephrine, which your brain releases when you are alert and which helps turn an experience into a lasting memory. The cells that supply it fire steadily while you are awake, slow in deep sleep, and go almost silent during REM. It hits the lowest level of your entire day at the exact time you are dreaming most.

J. Allan Hobson proposed that this is why dreams are so slippery: the chemistry that would normally pin an experience in place is turned down while the experience is happening. That is a named hypothesis, not a settled result. Memory researchers read the same quiet period the opposite way, as the window that lets the brain consolidate what it learned during the day. Either way, the practical picture holds. A dream arrives badly recorded.

What Ordinary Recall Looks Like

Know the baseline before you measure yourself against it. In 2025 a team led by Valentina Elce gave 204 adults a voice recorder and collected 2,900 morning reports, fifteen mornings each.

On 58% of mornings people could describe something. On 14% they were sure they had dreamt but could not say what happened, which researchers call a white dream. On 28% there was nothing. Roughly five dreams a week, then, with about one morning in four coming up empty. Blank mornings are normal and are not a sign the habit is failing.

The First Minute

Record it before you do anything else

This is the one habit with real support behind it, and everything else here is secondary. The methodological reviews are blunt: keep the gap between waking and reporting short, because both the length and the accuracy of the report drop as it grows.

In practice that means deciding the night before how you will record, and putting it where your hand already lands. Voice and text are each good at different things. In one study by Michael Schredl and colleagues, spoken reports came out about three times longer than written ones, but the writing condition produced more dreams reported overall and more connections drawn between the dream and waking life. Speak for detail. Write if you want to think about it later. The worst choice is choosing at 6am.

Record the fragment too. A room, a face, one feeling with no story attached. If 14% of your mornings are white dreams, a rule that only complete dreams count throws away most of your material.

An open notebook and pen, with small coloured fragments drifting up and away from the page
A notebook, a pen, and the pieces already leaving. Write the fragments down too: on 14% of mornings they are all you get.

The advice about waking up gently may be backwards

Almost every guide tells you to avoid a loud alarm and wake slowly. We printed that advice ourselves in an earlier version of this article. Then we tried to track down the study behind it.

What we found was a disagreement. A 2025 review of 69 awakening studies gives softer methods a modest win: being woken by someone calling your name rather than a buzzer added about 10 percentage points of recall, and awakenings the researchers called gentle added about 7. The standard methods paper in the field says the opposite outright, that abrupt awakenings favor recall over gradual ones. It points back to a 1963 study by Shapiro and colleagues, where people woken slowly more often thought they had merely been lying awake thinking. Waking gently blurred the line between the dream and the room.

Both sides woke people in the middle of the night, in a lab. Nobody has compared a loud alarm to a natural morning waking at home, which is the only version of the question you care about. So treat the soft-alarm advice as untested tradition. Try it, it costs nothing, but it is not why your recall is poor. The same goes for lying perfectly still with your eyes shut: no study of movement and dream recall exists. What is supported is narrower and duller. Do not spend that minute on something else.

Keep a log, with one honest warning

Writing dreams down does raise how often you remember them. Researchers have known it long enough that it now gets in the way of their own measurements.

Simply keeping a dream journal or being encouraged to recall dreams can dramatically increase dream recall frequency.

— Michael Schredl, interviewed in Discover, 2019

The warning is that it does not work the same way for everyone, and the exception is a strange one. Schredl’s own methodological work found that a dream diary raises recall in low and medium recallers and lowers it in people who already recall a lot. If you wake with two dreams most mornings, a journal may cost you some of them.

Give it longer than the internet tells you to. The one intervention we could find that measured this ran four weeks of daily morning ratings. Nothing supports the popular claim that recall improves in one to two weeks, so day nine is not a verdict.

A grid of small squares, most filled in terracotta, violet or lilac and some left empty, above a single ruled line
Weeks of entries with the blank mornings left in, which is what a log actually looks like after a month.

What Recall Is Not Evidence Of

Remembering more dreams does not mean you dream more, and remembering none does not mean you dream less. Recall and dreaming are two different things, and only one of them is under your influence.

The rare exception has been measured. In a sleep-lab population of 534 people, 6.5% said on a questionnaire that they never dream. Only 0.38%, about one in 250, held to it under interview, and across 36 laboratory awakenings of 16 such people, not one reported a dream. So true non-dreamers exist. They are very rare, and if you are reading a page about improving recall you are almost certainly not one. You are producing something like five dreams a week and losing most of them before you are properly upright.

That gap is what we built Dreamograph around: capture a dream by voice or text before you sit up, and keep every entry, fragments included, in a private history you can search later.

Three Questions People Ask Us

Why do some people remember so much more than I do?

The best current answer is about how you sleep, not how you dream. Researchers compared 19 high recallers with 17 low recallers in the lab and found the high recallers woke up more during the night: about 25 awakenings against 19, with more of them coming out of light non-REM sleep. Woken from that stage, high recallers gave a dream report 81% of the time, low recallers 30%. More brief trips to the surface, more chances to catch something on the way up. Mostly this is not a skill. Keep that in mind before you compare yourself to the friend who remembers everything.

Can a supplement help?

One decent trial exists. A hundred people took either 240 mg of vitamin B6 or a placebo right before bed for five nights, and the B6 group recalled significantly more dream content. No measurable effect on vividness, strangeness, color, or sleep itself. That dose is far above anything you get from food, so it is a question for a doctor rather than a shopping list.

How long until something changes?

Give it four weeks of one sentence a morning before you decide.

Sources