Baby Split Nights

A split night is a long stretch awake in the middle of the night — usually one to three hours — in a baby who is perfectly happy about it, and the most useful explanation is arithmetic rather than a sleep problem. There is roughly a fixed amount of sleep a baby of a given age tends to take in 24 hours. If too much of it is landing in the day, or bedtime is earlier than her body is ready for, the night cannot hold twelve unbroken hours — so it breaks in the middle instead of at the end. Nobody has published a word about this, which is the first thing worth knowing about it.

Not a syndrome, not a regression, not something she is doing to you. A day that no longer fits the night it is attached to, and the fix is in the day.

Nobody publishes this, and that matters

"Split night" is a term parents use. No health body this site is allowed to stand on uses it, defines it, measures how common it is or names a cause — not the NHS, not the American Academy of Pediatrics, not a paper on PubMed. Every confident explanation you will read for it, including the one on this page, is a mechanism somebody found plausible rather than a finding somebody measured.

What is published sits one level up and is worth having first, because it removes the alarming reading. Waking in the night is ordinary: the AAP describes a good sleeper at this age as a child who wakes frequently but can get himself back to sleep, rather than one who sleeps ten hours undisturbed, and calls frequent waking developmentally appropriate. The largest cohort in the subject followed more than 1,200 infants and found two patterns that never merged — about two thirds woke their mothers roughly one night a week throughout, while a third woke every night at six months, still twice a week at fifteen months and about once a week by two years. Night waking is not a stage that ends on a date.

So take everything below as a hypothesis you can test in a week rather than as a verdict on your night.

The arithmetic

Start from the one real number. The consensus figure is 12 to 16 hours of sleep per 24 for a baby of four to twelve months, naps counted inside the total rather than added to it, and 11 to 14 hours for a one- or two-year-old. Wide bands, but they have two ends, and the top end is not unlimited.

Now suppose your baby sits near the middle — say fourteen hours — and the day currently holds three of them in naps. That leaves eleven for the night. If she goes down at half past six and you are hoping she gets up at seven, you are asking a night to hold twelve and a half hours of sleep out of a body with eleven in it. The extra ninety minutes has to go somewhere. It usually goes in the middle: she completes a good long first stretch, surfaces at the point where the drive to sleep is lowest, finds she has nothing left to spend, and lies there chatting.

Two versions produce the same night. Too much day sleep for her age, or a bedtime earlier than her body is ready for. Both make the night longer than the sleep available to fill it, and both are things you can change by twenty minutes to see what happens.

Why it turns up when it does

Split nights cluster around the weeks when the day changes shape, and that is consistent with the arithmetic rather than a coincidence. A nap that has recently got longer, an unusually big afternoon sleep in the pram, a week of bedtimes crept earlier to cope with a bad patch, an illness that put extra daytime sleep into the day for a few days — each of these adds to the daytime side of the ledger while the night is expected to stay the same length.

They also turn up in the fortnight after a nap goes, for the opposite reason: bedtime moves earlier to absorb the change, sensibly, and then stays earlier after the day has caught up. That is worth checking if the split nights started shortly after a transition, and the by-age bands on the wake windows by age page will tell you whether the last stretch before bed is now short for her age.

The fixes, in order, one at a time

Change one thing and hold it for four or five days. Changing three at once teaches you nothing, and changing something new every second night leaves her with no pattern to learn. In the order most likely to work:

1. Trim the day sleep by twenty or thirty minutes, usually off the last nap of the day. Cap it rather than cutting it; a nap kept but shortened moves sleep to the night without emptying the afternoon.
2. Move bedtime fifteen to twenty minutes later. This is the opposite of the usual advice on this site, and it is the right direction here — the whole point is that the night is currently longer than the sleep available.
3. Check the last wake window. If the stretch between the final nap and bed is short for her age, she arrives at bedtime without enough pressure to sleep through to morning. The wake window calculator gives the band; the sleep schedule generator shows what the whole day looks like once you move one piece of it.
4. Then leave it alone for a fortnight. Days vary. What you are looking for is a run, not a good night.

You cannot judge any of this from memory at four in the morning, which is the argument for writing it down: bedtime, nap lengths, the time she woke and the time she went back. Five days on a printed day sheet usually shows the pattern before any change does.

One age exception: under about four months none of this applies, because sleep is not yet organised enough to be arranged. At that stage the night is broken because that is what a newborn night is.

The middle of the night, kept dull

While you are testing the day, the night has one job: to be boring. Dim, quiet, no lifting out unless she needs you, no lights and no conversation. Not absence — respond if she wants you — but nothing that makes two in the morning worth being awake for.

The reason is simple enough. A baby who is not distressed and not hungry is awake with time on her hands, and a parent arriving with a light on and a chat is the most interesting thing available. Boredom is doing real work here, and it works faster than it feels like it does.

Wherever she ends up in that stretch, the setup does not change: babies go down on their back, on a firm flat mattress, in a cot clear of blankets, pillows, bumpers and soft toys, in your room for at least the first six months. A baby awake and rolling around an empty cot for two hours is fine; a baby awake in a cot with things in it is a different situation, and the safe sleep guide covers why.

What this is not

It is not an early start. If she wakes at five and stays up, that is the end of the night rather than a hole in the middle of it, the causes are mostly the reverse of these, and the early morning waking page is the one you want — a later bedtime helps a split night and usually makes a five o'clock start worse.

It is also not distress. Everything here describes a baby who is cheerful: babbling, kicking, rolling around, occasionally complaining about the lack of company. A baby who is crying hard in the middle of the night, who cannot be comforted, who is inconsolable in a way that is new, is telling you something else, and no amount of trimming an afternoon nap will answer it.

When it is not arithmetic

Pain, illness, a baby feeding poorly or not gaining weight, a new cry that does not sound like her usual one, or a change you cannot account for is worth a call to your pediatrician, GP or health visitor — as is a worry that will not settle, and as is a household that has stopped coping with the nights. Bring your notes; a week of them says more than any description at an appointment. This page is general information, not medical advice, and it cannot see your baby.