Two to four months
2-4m · Updated 2026-08-23
Somewhere in these weeks, the clock switches on. The melatonin rhythm that simply didn't exist at birth becomes observable around nine to twelve weeks and steadies at thirteen to fifteen; body temperature and cortisol fall into daily rhythm; day and night become real categories for your baby, not just for you. This is the single fastest period of change in the whole first year — "the most rapid consolidation in infant sleep regulation occurs in the first four months" — and it changes what you can fairly expect, in stages, not all at once.
What arrives, in order
- Nights organize first. The longest sleep block plants itself in the evening and stretches. By three months, over half of babies sleep from midnight to 5 a.m. without interruption; by four months it's around seventy percent. If yours isn't there — that is a coin flip, not a failure.
- Bedtime migrates earlier. The drowsy time that used to arrive at 9–11 p.m. moves toward 6–8 p.m. across these weeks. Follow it down; an "early" bedtime here isn't a parenting choice so much as the biology relocating.
- The midmorning nap forms first — around twelve to sixteen weeks, near 9 a.m., initially a modest 40 minutes and growing. The midday nap follows a few weeks later. This order is fixed by the brain, not by your scheduling skill.
- The rest stays messy — on purpose. Nights can be genuinely good while naps are still brief, irregular catnaps. That combination is the expected state of three months, not a contradiction you caused.
What has not arrived yet
Wake windows are still short — roughly one to two hours, and that's a ceiling, not a target. Nap positions aren't stable enough to schedule by the clock; "the third nap" is not yet a real category. And around four months, a famous regression window opens — new sleep architecture, new wakings, one to two rough weeks for maybe half of babies. Knowing it is a maturation, not a setback, is most of surviving it.
I'll say the quiet part about my own model too: in 671 days of my child's logs, months zero to four were, by a wide margin, the hardest era to predict. The scatter you're living in is real, measured, and shared.
What the app does at this age
Between two and four months, NaniNani starts having opinions — gently. The tide becomes meaningful because sleep pressure itself is now fully online, so the water on the screen is worth a glance. But the app stays deliberately loose about clock times, because the clock underneath is still calibrating. It will not tell you the 9 a.m. nap is "late" while the 9 a.m. nap is still being invented.
The practical shape of these weeks: protect the short windows, chase the earlier bedtime as it appears, get bright light in the morning, and treat any scattered day as the construction noise of a clock being built. It is being built — on schedule.
Where this comes from
- Marc Weissbluth, Healthy Sleep Habits, Happy Child (4th ed.). Read in full; wake-window rules, turning points of sleep maturation, second-wind physiology.
- Alexis Dubief, Precious Little Sleep. Read in full; wake-time tables, the sleep-drive balloon, nap-transition mechanics.
- David Kennaway et al., melatonin ontogeny studies (1992, 1996). No diurnal melatonin rhythm in the first weeks; rhythm observable ~9–12 weeks, steady ~13–15 weeks.
- My own dataset: 671 days, 2,181 hand-logged sleep sessions of one child, month 1 to month 22, used to fit and validate the engine. The honest scorecard, losses included, is in the notebook.