Infant Sleep Regressions: A Guide to Every Stage

At some point in the first two years, usually several some points, your baby’s sleep will fall apart. They were sleeping reasonably. Then they weren’t. You type “sleep regression” into your phone at an hour no one should be awake and find seventeen contradictory articles.

Here is a map of what happens at each stage and why. Not every baby hits every regression, and the timing is approximate. But the developmental changes behind each disruption are well-documented, and knowing what is driving the problem helps you respond to it.

A word about the word “regression”

The term is informal. You will not find “sleep regression” as a clinical diagnosis in any pediatric textbook. What the research describes are developmental sleep disruptions: periods when the brain is building new capabilities and sleep quality temporarily suffers as a result [1].

The word “regression” is misleading in one important sense: it implies your baby is going backward. They are not. Their sleep is temporarily worse because their brain is doing something new and significant. The disruption is a side effect of progress, not a retreat from it.

The 4-month regression (3-5 months)

What is happening: Your baby’s sleep architecture is maturing. Newborns spend most of their sleep in deep stages; around 3-5 months, the brain reorganizes sleep into the cycling pattern adults use, moving through light NREM, deep NREM, and REM, with brief surfacings between cycles [1].

Why it disrupts sleep: Each surfacing between cycles is a potential full waking. Your baby may go from two night wakings to five or six. Before this shift, they could sleep through transitions between stages without noticing. Now each transition is a moment of partial arousal, and if the conditions that were present at sleep onset (your arms, a bottle, motion) are no longer there, the baby wakes fully.

How long it lasts: The architectural change is permanent. The disruption, meaning the period where your baby cannot yet link cycles independently, typically lasts 2-6 weeks [2]. The Henderson longitudinal study of 75 infants documented significant individual variation in how quickly babies learned to self-settle through these new cycle transitions.

What makes this one different: Every other regression is temporary. This one changes the structure. Sleep will improve, but it will not revert to the newborn pattern. This is the regression that makes sleep training relevant; before the architectural shift, there was no cycle-linking skill to teach.

The 8-10 month regression

What is happening: Three things at once. Separation anxiety, driven by maturing object permanence, means your baby now knows you exist when you are not in the room and minds about it. Motor milestones like crawling and pulling to stand get practiced in the crib. And many babies are transitioning from three naps to two [3].

Why it disrupts sleep: Separation anxiety makes bedtime and middle-of-night wakings harder. The baby does not just want comfort; they want you specifically, because they now understand that you are a persistent object who can be absent. Motor practice wakes the baby: pulling to standing and then being unable to sit back down is a common 2 am call. An awkward nap schedule leads to overtiredness, which makes everything worse.

How long it lasts: Typically 2-6 weeks, resolving as the milestones settle. The separation anxiety component tends to peak first and then moderate as the baby accumulates evidence that you always come back. The motor component resolves once the physical skill is solid.

What helps: Consistency above all. This is not the time to change the sleep approach. Practice the stuck positions (standing to sitting) during the day. Short separations with calm returns during waking hours build the baby’s understanding that leaving is not permanent.

The 12-month regression

What is happening: Walking (or the attempt), first words, and a growing desire for independence. Many babies also attempt to drop to one nap around this age, though most are not ready until 14-18 months.

Why it disrupts sleep: Walking practice in the crib at 3 am. Scher’s research on 52 infants found that motor milestone acquisition was directly associated with temporary sleep disruption; the brain rehearses new motor patterns during light sleep [3]. Increased separation protests at bedtime as cognitive development deepens the baby’s understanding of their own separateness. If the baby drops a nap too early, overtiredness.

How long it lasts: Usually 2-4 weeks. The nap issue may take longer to sort if the baby is fighting the second nap but not ready for one.

The biggest pitfall: Dropping to one nap too early. The Touchette study identified 12 months as a peak period of sleep fragmentation, and premature nap consolidation was a contributing factor [5]. Most babies are not ready for one nap until 14-18 months. A baby who refuses the second nap at 12 months is almost always going through the regression, not signaling genuine readiness.

The 18-month regression

What is happening: A language explosion, growing autonomy and the ability to say “no,” separation anxiety resurgence, and emerging molars.

Why it disrupts sleep: Toddlers at 18 months can and do resist bedtime. They can stand and call for you. They can throw things out of the crib. The combination of cognitive awareness and physical capability makes this regression feel different from the earlier ones: it is the first one where the child is an active participant in the disruption [5]. The Mindell study on sleep and social-emotional development found that the association between sleep disruption and developmental milestones was particularly strong during this period [4].

How long it lasts: 2-6 weeks, though the behavioral component (bedtime resistance) may persist longer if boundaries are not set. This is the regression where habits can form. If a toddler learns during these weeks that calling for a parent produces an extended visit, that pattern may outlast the developmental disruption.

What helps: Controlled choices (which pajamas, which book), a predictable routine, and clear limits delivered warmly. The routine provides structure; the choices provide agency within that structure.

The 2-year regression

What is happening: Imagination, nightmares, a full understanding of their own preferences, possible toilet training transitions, and the arrival (or anticipation) of a new sibling for many families.

Why it disrupts sleep: Nighttime fears, stalling tactics, requests for another glass of water, resistance to the crib. The 2-year-old can imagine things that are not present, including things that scare them. Nightmares become possible for the first time because the brain can now generate complex visual scenarios during REM sleep.

How long it lasts: Variable. The developmental changes are gradual rather than sudden, and the “regression” is less a discrete event and more a phase of testing boundaries. The acute disruption typically lasts 2-4 weeks; the behavioral patterns can persist if they become established.

What helps: Build common stalling requests (water, bathroom, one more hug) into the bedtime routine so they cannot be used as delay tactics. For nightmares, brief comfort and reassurance. For night terrors (deep-sleep episodes where the child screams but is not fully awake), stay close for safety but do not try to wake them.

Patterns across all regressions

A few things hold true across the board:

Consistency helps. The bedtime routine that was working before the regression is still the right routine. Changing everything in response to a disruption tends to extend the disruption [4]. The one adjustment worth making is bedtime: an earlier bedtime on days when naps are disrupted prevents the overtiredness spiral.

Not every bad night is a regression. Illness, teething, travel, schedule changes, and growth spurts all disrupt sleep. A regression is specifically tied to a developmental milestone and follows a recognizable pattern: sudden onset in a baby who was sleeping adequately, coinciding with a visible new skill or cognitive change.

They end. With the exception of the 4-month architectural shift (which is permanent but manageable), every regression resolves. The timeline varies, but the trajectory is toward improvement. The sleep your baby had before the regression is not gone. It is on pause while the brain finishes a project.

Your response shapes the aftermath. The regression itself is developmental and will pass regardless. What may persist is any new pattern introduced during the regression. A new sleep association (rocking to sleep, co-sleeping when it was not the prior arrangement, feeding to sleep) introduced during a regression may become the new expectation once the regression passes. If you are going to make a change, make it deliberately.

A note for tonight

If you are in the middle of one and cannot tell which one: it does not matter much tonight. What matters tonight is that your baby is safe, you are doing your best, and this specific night will end. The developmental explanation is for tomorrow, when you have had coffee. Tonight is just about getting through.

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.