Calling it a "regression" is misleading. Nothing is going backward. What's actually happening is that your baby's sleep is permanently maturing into something more adult-like, and the transition is rough.
If you're in the middle of it, this is probably the hardest stretch of the first year. It does end.
What's actually changing
For the first three months, babies sleep in a simple two-stage cycle and tend to drift between stages without fully waking. Around 12 to 16 weeks, sleep architecture matures. The cycles get shorter (about 45 to 60 minutes), and brief partial wakings appear at the end of each cycle. This is how adults sleep too. We just don't notice our own micro-wakes because we roll over and fall back asleep.
A baby who used to nap for two hours straight is now waking at the 45-minute mark. A baby who slept a 6-hour stretch at 10 weeks is now up every 90 minutes overnight. Nothing is broken. The brain is rewiring.
The reason it feels like a regression is that whatever helped your baby fall asleep before (rocking, feeding, bouncing) now has to happen at every cycle transition rather than once. If a baby falls asleep being rocked, and then wakes 45 minutes later in a flat crib, the conditions have changed and the brain calls for help.
When it tends to happen
Anywhere from about 3 to 5 months. Some babies hit it earlier, some later. Some babies who were "easy sleepers" get hit hard. Some babies who never slept well don't notice much of a change because they were already waking frequently.
The classic version: a baby who has been sleeping reasonably well for a few weeks suddenly starts taking 30-minute naps, fighting bedtime, and waking every 1 to 2 hours overnight. It feels like it appeared overnight. It usually did.
How long it lasts
Two to six weeks is the typical range. The change in sleep architecture itself is permanent, but the disruption is temporary. What's happening during those weeks is that your baby is learning to navigate the new, more complex sleep cycles. Once they figure out how to link cycles, things settle.
A baby who could not previously soothe themselves through a wake-up has to learn this skill, more or less, during this window. Some babies get there quickly on their own. Some need a bit of help. Some need a lot.
What genuinely helps
A few things that consistently make this stretch easier:
Tighten wake windows
At 4 months, most babies do best with wake windows of about 1.25 to 2 hours. A baby who's been pushing 2.5-hour wake windows because they "seem fine" often crashes into overtired bedtime. Pull the wake windows in by 15 to 30 minutes and naps tend to go in easier and last longer. (We have a full wake windows guide if you want the age-by-age numbers.)
Practice drowsy-but-awake at one sleep period a day
If your baby has only ever fallen asleep being rocked or fed, the new wake-ups will all need rocking or feeding too. You don't have to overhaul this. Pick one easy sleep period a day (often morning nap or bedtime) and put them down sleepy but still awake. Let them practice the last bit of the falling-asleep process themselves. It doesn't have to work every time.
Don't add new heavy sleep crutches mid-regression
This is the biggest practical mistake. A baby is suddenly waking every hour, the parent is exhausted, and a new habit gets established because it's the only thing that buys sleep tonight. Bedsharing started in week two of the regression. Bouncing on the yoga ball for 45 minutes at every wake-up. Feeding back to sleep every 90 minutes when feedings used to be every 4 hours.
If those things genuinely work for your family long-term, fine. But adding them in desperation tends to lock in patterns that are hard to undo at 8 or 10 months. If you wouldn't want it to be the answer in three months, try not to make it the answer this week.
Keep the room cool, dark, and boring
Sleep environment matters more now that brief wake-ups are happening. A baby who briefly opens their eyes in a pitch-dark, quiet room will often roll over and resettle. A baby who opens their eyes to a sliver of light, the dog moving, or the heater clicking on may come fully awake.
White noise (continuous, not too loud) and full blackout help. They help a lot.
Track what's actually happening
In the middle of a tough stretch, every night feels identical. It often isn't. If you log feeds and naps in Tottli during the regression, you'll usually see a pattern emerge after 4 or 5 days, often something like "third nap is too late and bedtime collapses" or "the longest stretch has actually grown by 30 minutes since last week." Data beats memory when you're sleep-deprived.
What not to do
A few things that tend to backfire:
Starting major sleep training in the middle of the regression. If your baby is in the thick of it, formal sleep training (extinction, Ferber, etc.) is unusually rough on everyone. Most sleep consultants will tell you to wait until the disruption has settled before starting a structured program. There's no rule, but trying to teach a new skill while the underlying system is in flux is usually harder than waiting two weeks.
Assuming patience alone will fix it. "It's a phase, just wait it out" is half right. The architecture change does pass. But if you've been feeding to sleep at every wake-up for three weeks, that pattern won't dissolve on its own. Some change is usually part of the path through. Patience plus small adjustments tends to beat patience alone.
Capping naps to "make" night sleep better. A common impulse is to wake a baby from a long nap to protect bedtime. At 4 months, the more common problem is naps that are too short, not too long. An overtired baby sleeps worse at night, not better. Let them sleep.
Comparing to other babies. This is the wrong age to do it. Some 4-month-olds are sleeping 8 hours a night. Some are up every hour. The variation is enormous and not predictive of how things look at 12 months.
When to be patient vs when to change something
The honest answer is that both can be right depending on the situation.
Patience is reasonable when:
- Things are gradually getting better, even slowly
- Your baby is healthy, gaining weight, and seems happy when awake
- The disruption is in week 1 or 2 and you can stretch a little longer
Something needs to change when:
- You're four-plus weeks in with no improvement
- Bedtime takes more than 45 minutes most nights
- You're at the edge of what you can sustain physically or emotionally
- A new sleep crutch has appeared that you don't want to keep
There's no medal for white-knuckling a pattern that isn't working. Adjusting wake windows, dimming bedtime cues, working on independent falling-asleep at one sleep period a day, those are reasonable changes that don't require a full sleep training program.
A reframe
The 4-month regression is a developmental milestone, not a problem to solve. Your baby's brain is reorganizing how sleep works, and the new system is more sophisticated than the old one. The work in front of you isn't fixing a regression. It's helping your baby learn the skill the new system requires, which is linking sleep cycles without fully waking up.
Some babies get there in two weeks. Some take six. Almost none take three months. If you're in the middle, the thing to remember is that the chaos is temporary even though the change is permanent, and the version of sleep that comes out the other side is genuinely better than what came before. It just doesn't feel that way yet.