Sleep regressions and how Ferber handles them

Every well-trained sleeper has bad weeks. The bad weeks usually have names: the 8-month regression, the 12-month regression, the 18-month regression. This guide is what to do during one.

What a regression actually is

A regression is a sudden, temporary disruption of a sleep pattern that was previously stable. It is almost always tied to a developmental change: crawling, pulling to stand, walking, language burst, separation anxiety spike. The baby is busier inside, so the night is busier too.

It is not a failure of training. The Sleep Foundation describes the same pattern: a baby who slept through reverts to one or two short wakings or to longer time-to-asleep at bedtime, then returns to normal within a couple of weeks.

The job during a regression is twofold: do not undo the protocol, and do not add new sleep crutches that will need their own un-training later.

The 8-month regression

Hits between roughly 7 and 10 months. Drivers are usually:

• Crawling. Babies practice in bed.
• Separation anxiety, first wave. Drop-off cries return briefly.
• Stronger object permanence. Your baby can now think “the parent is in the other room” instead of “the parent does not currently exist.” This is good cognition and bad sleep.

What it looks like: longer time-to-asleep, one or two new mid-night wakings, sometimes early-morning waking by an hour.

The fix: a short Ferber re-arc. Most families do not need to start from Night 1. Start at the night that matches your baby’s current time-to-asleep. If your baby took ten minutes to fall asleep before the regression and is now taking twenty, start at Night 2 or 3. Run two or three nights at that level and the pattern usually returns.

The 12-month regression

Hits between roughly 11 and 14 months. Drivers are:

• Walking. Same energy as crawling, applied to a more demanding skill.
• Nap transition. Some babies are dropping the third or second nap around now. The day’s sleep architecture is changing.
• Language burst. Your baby is rehearsing words at bedtime, which delays drop-off.

What it looks like: bedtime fights, longer time-to-asleep, an early waking that lands consistently around 5 a.m.

The fix: a Ferber re-arc plus a daytime adjustment. Cap any nap that ends after 3 p.m. Move bedtime ten to fifteen minutes later if the early morning waking persists. Run two or three Ferber nights to reset the bedtime drop-off. The Huckleberry team recommends the same nap-and-bedtime tweak.

The 18-month regression

Hits between roughly 16 and 20 months. Drivers are:

• Big language and identity development. “No” becomes a tool. Bedtime is the easiest place to use it.
• Separation anxiety, second wave. Sometimes intense.
• The transition out of the second nap, for late nappers.

What it looks like: outright refusal to go to bed, multiple stalling tactics during the routine, mid-night wakings with calls for parents.

The fix is more behavioral than the earlier two. Hold the routine firm. Address the requests calmly and briefly: one drink, one bathroom trip, one last cuddle, then the door closes. Run a short Ferber re-arc if drop-off crying returns. Some families find the toddler-version intervals work better at this age: they start at five, ten, fifteen rather than three, five, ten.

How to do a short re-arc

The same protocol, fewer nights. Most regression re-arcs take two to four nights, not seven, because your baby already knows what a Ferber wait looks like.

1. Pick the night that matches your baby’s current time-to-asleep. Use the table from the 7-night arc guide as the reference.
2. Run that night’s intervals tonight.
3. Tomorrow night, step up one row in the table.
4. Step up again the next night.
5. Once you are at the night that matches your baseline, hold there for a night, then resume normal life.

If by Night 3 of the re-arc you are not seeing improvement, run the diagnostic from the hard-middle guide. Sometimes a regression is masking an illness or a tooth that is not visible yet.

What not to do during a regression

The regression is the moment families undo a year of training. The classic mistakes:

• Bringing the baby into bed. Once. The baby learns that the way to end the regression is to call for it. Now the regression is permanent.
• Adding a new feeding. A baby who was not feeding overnight does not need a new feed. They need a check-in and the protocol.
• Lengthening the bedtime routine to compensate. A 45-minute routine becomes the new minimum. Hold the routine to its previous length.
• Quitting and trying a different method. A regression is the worst possible time to switch methods. The training the baby already has is the fastest path back to a sleeping baby.

When to involve a pediatrician

If a regression has lasted more than three weeks with the protocol running cleanly, that is not a regression any more. Talk to your pediatrician. Common findings at that point are an ear infection, a tooth that is not visibly through, a feeding pattern that has shifted (breastfeeding mothers sometimes have a milk-supply dip), or sleep apnea in older babies with enlarged tonsils.

What the app does for regressions

The history view is most useful at this point in your baby’s life. A regression will show as a clear bump on the time-to-asleep chart, which helps separate “actually a regression” from “this week felt rough but the data is fine.”

The app’s re-arc flow lets you start from any night, so a 12-month regression does not need a Night 1. You pick where to start, the intervals load, and the timer runs.