← the guide

What moves and what holds

every age · Updated 2026-08-26

Over the first two years, almost every sleep number you can watch will travel a long way: bedtime will fall by about three hours and then climb most of the way back, naps will go from four or five a day to one, and day sleep will shrink from six hours to two. One number holds still through all of it — the night. Knowing in advance which numbers are travelers and which one is the anchor spares you a great deal of worry, because the travelers look like problems and aren't.

The travelers

Bedtime is found, not chosen. It lands where the last sleep of the day ends plus the longest stretch your child can stay awake — and those two move at different speeds. In the first half-year the last nap retreats faster than the wake window grows, so bedtime falls, in my son's data from around 10 p.m. at one month to 7 p.m. near six months. Then the last nap stops retreating, the window keeps growing, and bedtime climbs back toward 9 p.m. by age two. Both directions are the same arithmetic working itself out. A bedtime that is drifting is not a routine failing; it is a child developing on schedule.

Naps leave in steps, and the steps have messy edges. Four-ish naps, then three, then two, then one — but each hand-off takes a month or two of mixed days, and a two-nap Tuesday in a one-nap week is the transition working, not breaking (the second year walks the biggest one). Between transitions, expect surprising rigidity: my son once held exactly two naps a day for two straight months, then later exactly one a day for most of a year. Steps, not slides.

Day sleep shrinks — and that is need declining, not sleep escaping. Six hours of day sleep at one month, about two by eighteen months, still about two at two years. The surviving midday nap holds near two hours and drifts a few minutes a month later, pushed by the same growing wake window that is pushing bedtime.

The anchor

From roughly five or six months, the night settles at about ten measured hours — wakings subtracted — and stays there. In my son's logs, every monthly median for twenty-two straight months landed between 9h17 and 10h57, while the day above it went from three naps to two to one. The only visible wobble was during the two-to-one nap transition, and it recovered by itself within a month. (Reference charts like the Zurich percentiles will say eleven hours at these ages; they count time in bed, so your own logged night should read lower.)

The practical meaning: when the day changes shape — a nap dropped, a bedtime sliding somewhere new, a day total that keeps shrinking — the night is not in danger. In two years of data it absorbed every one of those changes without giving anything up. The reflex to defend the night against the day's reorganizations is understandable, sold to constantly, and unnecessary. The measured version of this, losses and caveats included, is in the notebook: everything moved except the night.

What to do with this

Mostly nothing — that is the point of an expectations article. The travelers travel on their own schedule and the anchor holds on its own. The two useful actions are the ones the per-age articles already teach: during a nap transition, spend an earlier bedtime freely (every dropped nap anywhere in childhood is paid for at bedtime), and while any single day looks chaotic, judge the week instead. Your child's cues outrank every curve on this page.

Where this comes from