The 4-Month Sleep Regression: What Is Actually Happening and What You Can Do

2:47 am. Second wake in three hours. You have fed them, changed them, rocked them. They are awake again, and you are sitting with them in the dark calculating how many hours until morning.

If your baby is somewhere between three and five months old, what is happening has a name: the 4-month sleep regression. The name is misleading. It is not a regression. What is happening is a step forward, and understanding the difference changes everything about how you respond to it.

What is actually changing

In the first few months of life, your baby had only two sleep states: active sleep (similar to REM) and quiet sleep (similar to deep NREM). They moved between these two states in short cycles, and they spent a disproportionate amount of time in the quiet, deep phase. That is why newborns can fall asleep anywhere and stay asleep through noise, light and movement. Their sleep was simple and heavy, and the transitions between states were brief enough that most babies slept right through them.

Around 3-5 months, the brain matures. Sleep architecture reorganizes into the same multi-stage pattern adults use: cycling through light NREM stages (N1 and N2), deeper NREM sleep (N3), and REM sleep, with a brief surfacing at the top of each cycle [1]. For adults, these surfacings are so quick we do not remember them. For a baby who has never experienced them before, each one is a full waking. The baby opens their eyes, registers that the environment exists, and does not yet have the skill to close them again and drift back under.

Your baby is now cycling every 45-60 minutes instead of sleeping in long consolidated blocks [2]. The deep anchor sleep of the newborn phase is gone. It is not coming back. That is the first piece of news.

The second piece of news is that this is normal and necessary. The mature sleep pattern your baby just acquired is the same one they will use for the rest of their life. It allows for more efficient memory consolidation, better regulation of growth hormone release during deep sleep, and the kind of restorative cycling that supports brain development. The problem is not the new architecture. The problem is that your baby does not yet know how to navigate it.

Why this is not like the other regressions

The 8-month, 12-month and 18-month disruptions are temporary. They are tied to specific developmental milestones: separation anxiety, standing, language, the cognitive leap that lets a toddler understand that you exist in another room. When the milestone settles, so does the sleep. These are software updates, if you want to think of it that way. The baby’s operating system gets a new feature, runs rough for a few weeks while it integrates, and then smooths out.

The 4-month shift is a hardware change. Your baby’s sleep has permanently changed its architecture. The cycling, the surfacings, the lighter sleep in the second half of the night: these are features of the new system, not bugs that will resolve.

This sounds worse than it is. The change is permanent, but the disruption is not. What is temporary is your baby’s inability to link those new sleep cycles together independently. That is the skill they need to learn, and most babies can learn it within 2-6 weeks [3]. In the Henderson et al. longitudinal study of 75 infants, the majority were able to self-settle and sleep in consolidated stretches by 5 months, though there was significant individual variation. Some babies figured it out in days; others took considerably longer.

What actually helps

There is no way to reverse the change, and anyone suggesting you can is selling something. But there are things that support your baby through it.

A consistent bedtime routine. A predictable sequence of events (bath, feed, book, song, bed) signals to the developing circadian system that sleep is coming. One large study of 405 mothers found that a consistent nightly routine reduced night wakings and improved total sleep time within three weeks [1]. The routine works because repetition builds an associative chain: each step tells the brain that sleep is approaching, and over time the routine itself begins to lower arousal before the baby even reaches the crib.

Separating the last feed from sleep onset. If your baby falls asleep while feeding and then wakes at the top of a cycle, they look for the conditions that were present when they fell asleep. The feed is not there. They cry. Moving the last feed earlier in the routine, so the baby goes into the crib drowsy but awake, helps them learn to fall asleep under the same conditions they will find at 2 am.

The sleep environment. A dark room, a cool temperature (between 68 and 72 degrees Fahrenheit), and a consistent sound environment all support longer stretches of sleep. Darkness matters because even small amounts of light can signal the developing circadian system that it is time to wake. Cool air matters because core body temperature naturally drops during sleep, and a warm room works against that process. White noise can help mask the household sounds that are more likely to trigger a wake during the lighter sleep stages your baby now spends more time in.

Allowing some space before responding. This does not mean cry-it-out and it does not mean ignoring your baby. It means pausing for a minute before rushing in, because some of those cycle-top wakings will resolve on their own if given the chance. Some will not. You will learn your baby’s cries and which ones settle. The brief fusses that last 30 seconds and trail off are often the baby practicing the skill of returning to sleep; the escalating cries that build in intensity are a genuine request for help.

Sleep training at four months

If the disruption persists beyond a few weeks and you are considering a more structured approach, the evidence supports several methods. Graduated extinction (Ferber), full extinction, and bedtime fading have all been shown to reduce night wakings, with no measurable differences in stress hormones or attachment at follow-up [4].

The timing question is worth considering. Some sleep researchers suggest waiting until the worst of the disruption has passed, usually 1-2 weeks after the regression begins, before starting a formal method. The reasoning is practical: during the peak of the regression, it is hard to distinguish between developmental waking (the baby adjusting to new sleep architecture) and habitual waking (the baby needing a specific condition to fall back asleep). Starting sleep training during the peak can mean working against two problems at once.

That said, there is no strict minimum age set by the AAP for behavioral sleep interventions, and many families begin at 4 months with good results. If you do start during the regression, choose a method and commit to it for at least a week before evaluating. Switching methods resets the learning process and extends the disruption for everyone.

The choice of method is a family decision. The site’s position is that the evidence supports multiple approaches, and the best one for your family is the one you can do consistently.

What to expect

The worst of the disruption usually peaks in the first 1-2 weeks after sleep architecture matures, then gradually improves. Some babies adjust within days; others take 4-6 weeks. If you are implementing a consistent routine without formal sleep training, expect gradual improvement over 3-6 weeks. Progress is not linear: you will have a good night followed by a terrible one and wonder whether you imagined the improvement. You did not. The trend matters more than any single night.

If you choose a sleep training method, most research shows significant improvement within 3-7 nights for graduated extinction [4]. Full adjustment may take 1-2 weeks. Bedtime fading takes longer, typically 2-3 weeks, but involves less crying.

One thing to watch for: if the regression coincides with the baby learning to roll, you may see additional disruption as the baby practices the new motor skill at night. This is temporary and usually resolves within a week or two of the baby becoming comfortable with rolling in both directions.

A note for tonight

If you are reading this at 3 am, here is what I can tell you: this is the hardest part. Not of the regression, necessarily, but of the uncertainty. You do not know yet how long this will last, which approach will work for your baby, or whether the advice you are reading will apply to your specific child. That uncertainty, combined with severe sleep deprivation, is its own kind of difficult.

Your baby is not broken. Their sleep has grown up faster than either of you were ready for. What you are doing tonight, whatever it is, whether that is rocking them or letting them settle or feeding them back to sleep, is surviving. That is enough for tonight. The developmental change that is making this hard is also the change that will eventually make longer, better sleep possible. You are in the difficult middle part, and it does end.

If you’re concerned about your baby’s sleep or health, speak with your pediatrician or family doctor. The American Academy of Pediatrics has sleep safety guidance at healthychildren.org. If your own sleep deprivation is affecting your daily functioning or mental health, that is also worth raising with your doctor.