The dream feed is one of the few baby-sleep techniques with an actual randomised trial behind it. It is also routinely oversold, because the trial was small, ran in 1993, and tested a bundle of three things rather than the one thing everyone quotes. Both of those are true at once, and knowing which part is which is most of what makes the technique useful.
What a dream feed actually is
A feed you give — breast or bottle — while your baby is still asleep or barely rousing, usually somewhere between 10pm and midnight, before you go to bed yourself. You are not waking them properly and you are not waiting for them to ask. The idea is to put milk in before the first night waking would have happened, so the longest stretch of sleep lands while you are also asleep.
The one trial, and what it found
Twenty-six first-time parents and their newborns were randomised, thirteen to each group. The treatment parents were told to offer a "focal feed" between 10pm and midnight every night, to gradually lengthen the gaps between middle-of-the-night feeds using other soothing first, and to maximise the difference between day and night in light and handling.[1]
100% vs 23%
By three weeks the treatment infants had significantly longer night-time sleep episodes, and by eight weeks all thirteen were sleeping from midnight to 5am, against three of the thirteen controls. Crucially, total milk intake over 24 hours did not differ between the groups — the babies made up the night-time gap by taking more in the early morning.[1]
What it is reasonable to expect
The mechanism is redistribution, not reduction. Babies of this age need a certain volume across 24 hours, and where feed frequency has been measured directly it varies enormously between babies while the daily total stays fairly stable.[2] A dream feed moves milk earlier in the night; it does not persuade a baby to need less of it.
- A realistic win: the first waking moves from 1am to 3am, so you get one solid block instead of two broken ones.
- Not a realistic win: dropping night feeds altogether in a baby who still needs them.
- It tends to help most in the first few months, which is the only period the trial covers.
- It costs nothing to try for a week, and a week is long enough to see whether it does anything for you.
How to do one
The dream feed, in practice
Keep it boring. The whole point is that this feed does not become an event your baby wakes up for.
- 1
Pick a time and hold it
Somewhere between 10pm and midnight, and ideally before your own bedtime — a dream feed you have to set an alarm for after falling asleep costs more than it returns.
- 2
Lift gently, do not properly wake
Enough rousing to feed safely and no more. Keep the lights low and the room quiet; the day/night contrast was part of the trial's package.
- 3
Offer breast or bottle as usual
Some babies latch straight away; some need a stroke of the lip or a nappy change first. If your baby will not feed after a few minutes, put them down and try a different time tomorrow.
- 4
Wind and settle as normal
Hold upright for a few minutes if they usually need it, then back into their own sleep space, on their back, before you go to bed.
- 5
Give it a week before you judge
Track when the first waking happens, not how you felt in the morning. Two nights tells you nothing; seven tells you plenty.
Safety, which is not optional
A sleepy feed is exactly the situation where safe-sleep rules matter most. Feed your baby out of any sling or carrier, never prop a bottle, and put your baby back into their own clear, flat sleep space on their back afterwards.[5] If you know you are likely to fall asleep during the feed, make the surface you are on as safe as it can be before you start rather than after.[4]
When it backfires
- Your baby wakes fully and then treats 11pm as the start of a party.
- The dream feed replaces a feed they would have taken anyway, and the 3am waking arrives on schedule regardless.
- It becomes an alarm-clock habit you keep long after your baby stopped needing it.
- It masks a wake-window problem: if the last stretch of the day is wrong, no feed will fix the night. Wake window stacking is the thing to check first.
How it differs from the other night feeds
| Who initiates | Typical timing | What it is for | |
|---|---|---|---|
| Dream feed | You | 10pm – midnight | Moving the long stretch into your own sleep |
| Night feed | Baby | Any time | Meeting a genuine need for milk |
| Cluster feed | Baby | Late afternoon to evening | Loading up before the longest stretch |
When to drop it
When it stops changing anything, or when it starts causing a waking rather than preventing one. Try skipping it for three nights and compare the time of the first waking. If nothing gets worse, it has done its job. If your baby is waking for an hour in the middle of the night whether or not you dream feed, that is a different problem — see split nights.
And judge the whole thing on the 24-hour total rather than on any one night. The AASM consensus recommends 12–16 hours of sleep per 24 hours for infants aged 4–12 months, including naps, and says nothing at all about how that should be divided up.[3] A dream feed is a scheduling tool for your night, not a target your baby has to hit.
