Sleep regression ages: 4, 8, 12 and 18 months, explained
Updated 24 August 2026 · by Josh, Pipkin's developer
Sleep regressions cluster around predictable ages, roughly 4, 8, 12 and 18 months, each driven by something real: sleep-cycle maturation, gross-motor leaps, separation anxiety, molars, independence. Most last 2–6 weeks. Taking each window in turn: what drives it, and how to tell a true regression from an off week.
The regression windows at a glance
| Age | What's driving it | Typical length |
|---|---|---|
| ~4 months | Sleep cycles mature into adult structure (permanent) | 2–6 weeks |
| 8–10 months | Crawling, pulling to stand, separation anxiety, 3→2 nap transition | 2–6 weeks |
| ~12 months | Walking, language burst, first flirtation with dropping a nap | 1–4 weeks |
| ~18 months | Independence ("no!"), molars, separation anxiety round two | 2–6 weeks |
Some babies also wobble around 6 months (growth spurt, solids starting) and 2 years (nightmares, big-bed moves, potty training). None of these windows is guaranteed. Plenty of babies skip one entirely, and a regression that arrives a month "late" is still on time for your baby.
4 months: the one that changes everything
The only regression that's a permanent neurological change rather than a phase: sleep reorganises into light-first, ~45-minute cycles. It gets its own full guide.
8–10 months: the busy-brain regression
Your baby just learned to crawl or pull to stand, and the cot is now a gym. Add separation anxiety, newly able to miss you, plus the 3-to-2 nap transition, and sleep gets noisy. What helps: daytime practice of the new skill (it burns the urge), rock-solid bedtime routine, and holding two decent naps while the third scrappy one falls away.
12 months: the false nap-drop
Walking and a language burst can disrupt nights briefly, and many one-year-olds suddenly refuse a nap, which parents read as readiness for one nap. It usually isn't: most toddlers aren't ready until 13–18 months. Hold the two-nap structure for a couple of weeks before deciding. The 2-to-1 transition guide covers the real readiness signs.
18 months: the boundary-testing regression
A toddler who can say "no", walk away, and genuinely prefer your company to sleep. Molars often gatecrash the same window. What helps: consistent, boring responses at night, a bedtime routine they can predict to the second, and daytime choices ("red pyjamas or blue?") that spend their independence budget somewhere harmless.
Regression or just a bad week?
Illness, teething, travel and a too-late bedtime all impersonate regressions. The tell is the shape: a true regression is a sustained shift against your baby's own baseline across a week or more, not two rough nights that trace back to a snotty nose. This is where a log beats memory: sleep-deprived brains overweight last night and forget last Tuesday entirely.
In Pipkin
Pipkin watches your baby's log against their own baseline and names a regression window only when the data confirms it (wakings up, naps fragmenting, settling stretching), then shows the reasoning and what parents often try. Illness you've logged (teething, a fever, travel) gets factored in as the likelier explanation. Trends must hold for three weeks before Pipkin calls them, so a rough fortnight never gets promoted to a pattern.
Quick answers
What ages are sleep regressions?
The common windows are around 4 months (sleep-cycle maturation), 8–10 months (crawling and separation anxiety), 12 months (walking and nap pressure) and 18 months (independence and molars). Some babies also wobble at 6 months and 2 years, and some skip windows entirely.
How long does a sleep regression last?
Most last 2 to 6 weeks. If disrupted sleep runs past 6 weeks, it's worth checking for something sustaining it: a new sleep association, a schedule that no longer fits, or a medical cause like ear infections.
Is it a sleep regression or teething?
Teething disrupts sleep for a few days around a tooth's arrival, usually with daytime clues (drooling, chewing, one flushed cheek). A regression is a sustained multi-week shift in the whole sleep pattern. A log makes the difference obvious. Two bad nights is not a regression.
A note on medical advice: this guide is general information, not medical advice, and Pipkin is a tracking tool, not a medical device. Every baby is different. For anything that worries you, talk to your paediatrician, health visitor or midwife.
