"Six months" is the number every new parent has heard. The real picture is messier and more useful: four distinct night-waking trajectories, a regression cohort that is larger than the popular literature suggests, and an inflection point that sits closer to month 7 than month 6.
The phrase "sleeping through the night" is used in three different ways in the pediatric literature, and the differences matter for interpreting any cohort. The narrowest definition is the one Anders proposed in the 1970s — eight uninterrupted hours of sleep between midnight and 8 a.m. The intermediate definition, popularized by Henderson et al. in their seminal 2010 cohort of 75 New Zealand infants, is a five-hour stretch between 10 p.m. and 6 a.m.4. The broadest is the parent-reported version — the night a parent says "my baby slept through" — which is more permissive and often reports earlier consolidation than the objective measures.
For this analysis we use the Henderson five-hour-stretch threshold. It is the most widely cited objective measure in pediatric sleep research, aligns with the American Academy of Sleep Medicine's pediatric sleep consensus3, and tolerates the early-evening "split night" in which many breastfed infants get one wake-up between 10 p.m. and the early-morning hours without disrupting the longer block.
Operationally: a Wermom user is counted as having reached "first five-hour stretch night" on the first calendar night with no parent-logged feed or comfort event in any rolling five-hour window between 10 p.m. and 6 a.m. We then track whether that night becomes the new norm (≥4 of the next 7 nights also meet the threshold) versus an isolated good night.
Across the full cohort, the median age at which an infant achieved the rolling-norm five-hour stretch was week 28 — meaning month 7, not month 6. The 25th-to-75th-percentile band was wide: 25th percentile at week 21 (just under 5 months), 75th percentile at week 35 (just past 8 months). By the AAP's frequently quoted "by 6 months, many infants are capable of sleeping for longer stretches"1, the curve is consistent: many, yes — but not most.
This finding aligns more closely with Henderson et al., 2010, who reported a median of 12 weeks for the first five-hour stretch ever but a median of 26 weeks for the equivalent of the "rolling norm" measure4. Where the popular guidance overshoots is the difference between "did it once" and "does it consistently." Roughly 78% of Wermom-tracked infants logged at least one five-hour stretch by week 22, but only about half of those continued the pattern into the following two weeks. Pediatricians can use the rolling-norm distinction to set expectations with families: a great night at 4 months does not predict a great month at 4 months.
The plurality of infants in this cohort followed a fairly smooth, gradient-style consolidation. From a baseline of 3.4 mean wakings per night at week 16, the linear group dropped by roughly 0.3 wakings per week, crossing the 1-waking threshold by week 27 and the five-hour-stretch rolling norm a week or two later.
Two features stood out in this group. First, their consolidation timing was insensitive to the developmental "4-month regression" — there is no detectable bump in their week 17–19 wakings beyond what is attributable to noise. Second, their parents tended to log feeding patterns that were already largely day-loaded by week 16, with more than 80% of total daily intake before 9 p.m.
This is not a causal claim. It is consistent with the broader sleep literature in which behavioral consolidation tends to lag day/night caloric reorganization by 2–4 weeks5. For the families who want a single metric to track, the day-loaded feeding fraction at month 4 is at least a directional indicator of which trajectory an infant may be on. For parents using the Wermom App's feeding tracker, the day/night intake split is one of the simplest derived metrics to monitor between weeks 14 and 18.
Roughly 3 in 10 infants showed two distinct drop-offs separated by a plateau of 6–10 weeks. The first drop typically appeared in weeks 18–20 — coinciding with the developmental window in which sleep architecture matures from the more diffuse newborn pattern into adult-like sleep stage organization3. The second drop appeared in weeks 28–32, often after the onset of stable supported sitting and the introduction of complementary foods.
The plateau between is the part that frustrates parents and prompts most "Why won't my baby sleep?" pediatrician visits. From a clinical standpoint, the plateau is not a failure of sleep training; it is a normal feature of a multi-pathway developmental process. Pediatricians who see infants in this trajectory at the 6-month visit are often seeing an infant who has reorganized once and is awaiting the second reorganization.
About one in five infants in the cohort showed only a modest reduction in night wakings until weeks 32–36, after which they consolidated relatively quickly. Their week-by-week curves can be discouraging for parents because nothing visibly improves between months 4 and 7, and then a sharper change appears in months 8–9.
Within this group, two sub-patterns are clinically worth noting. The first is bedshare-associated: parents who reported a regular contact-sleep or bedsharing pattern were over-represented in this trajectory, consistent with prior work showing more frequent — but shorter — wakings in contact-sleeping infants4. This is not a value judgment about bedsharing. The AAP's 2022 task force is clear that the safest sleep surface is a separate, firm, flat, non-inclined surface in the parents' room for the first 6 months2, and Wermom's editorial guidance follows that recommendation. The trajectory data simply documents that the consolidation curve for these families differs from independent-sleeping families and shifts later, not that anything is "wrong."
The second sub-pattern in the late-consolidator group is feeding-driven: infants whose parents logged night feeds for nutritional reasons (preterm origin, low weight-for-age trajectory at the 4-month visit) clustered in this group. For these families, late consolidation is appropriate and the goal is not to compress night feeds prematurely. Pediatrician oversight matters more than population averages.
Roughly 1 in 8 infants in the cohort showed a 2–4 week increase in nightly wakings centered on weeks 17–19, after a period of improvement. The popular literature calls this the "four-month regression."6 In the Wermom data, the regression cohort's mean wakings increased from 1.9 per night at week 15 to 2.7 at week 18 before retreating to 1.6 by week 22. The duration was modal at 18 nights, with an interquartile range of 12–25 nights.
This is consistent with the leading developmental account: between months 3 and 5, infants transition from the relatively undifferentiated active/quiet sleep states of the newborn brain to a more adult-like organization with discrete REM/NREM cycles. Brief arousals between cycles are normal, and infants who had previously slept through them often briefly require assistance to do so — until they consolidate the new architecture3.
For pediatricians, the actionable distinction in the 4-month visit is between a regresser and a previously-poor-sleeper. The regresser pattern is self-limited, almost always resolves within 4 weeks, and does not warrant intervention beyond reinforcement of safe sleep and consistent wind-down2. Persistent sleep disruption at month 5 or beyond, by contrast, may warrant exploration of nighttime feeding patterns, sleep environment factors (ambient light, room temperature, sound), and — if relevant — discussion of behavioral interventions5. For families navigating the regression, the parent-facing companion guide on evidence-based sleep practice and the more detailed 4-month regression deep-dive may be useful.
This was not a sleep-training trial. The cohort included families using a wide range of approaches, from no behavioral changes at all to parent-directed graduated extinction to chair methods and check-and-console. We did not randomize, we did not collect granular method-fidelity data, and we did not control for selection (parents who actively track sleep in an app are likely more engaged with sleep generally).
What the data is consistent with is the Mindell et al. 2014 meta-analysis of behavioral sleep interventions, which found that across 52 studies, behavioral interventions improved sleep outcomes in about 80% of infants, with sustained effects through 1 year and no documented adverse effects on attachment or emotional development5. The 2018 AAP-endorsed clinical report on the management of bedtime problems and night wakings in young children echoes this: behavioral approaches are safe, effective, and consistent with infant attachment when implemented in a supportive context1.
For families who choose a behavioral approach, the trajectory data above suggests the strongest predictive window for intervention onset is weeks 18–20 (after the 4-month regression resolves) or weeks 26–28 (the second consolidation window for stair-step infants). Earlier interventions, while not unsafe, often pull against the underlying neurobiology of sleep maturation and require longer to take effect.
Several limitations deserve careful attention before applying these findings clinically.
Self-selection. Wermom App users are not a representative US infant population. They skew toward parents with higher digital engagement, more day-to-day tracking interest, and — based on app analytics — geographically toward urban areas. Sleep consolidation timing may differ in less-tracked populations.
Definition sensitivity. The five-hour-stretch threshold is one of several reasonable operationalizations of "sleeping through the night." Choosing the Anders eight-hour threshold shifts the population median later by approximately 6 weeks; choosing a parent-reported threshold shifts it earlier by approximately 3 weeks. The trajectory shapes are robust to this choice; the medians are not.
Causal inference. This is descriptive epidemiology, not a trial. We do not infer that any feeding pattern, sleep arrangement, or behavioral approach causes any of the trajectories. We do think the trajectories themselves are clinically useful as a normalization framework — "your baby is on a trajectory" is more accurate and more reassuring than "your baby is behind."
Future work. We plan to release a follow-on analysis pairing this cohort with parent-reported postpartum mood data (using validated instruments such as the EPDS where consent was given) to characterize how trajectory shape correlates with parental mental health load over months 4–9 — a question with substantial clinical importance that the current literature handles only superficially. That work is in pre-registration as of May 2026.
The Wermom App's sleep tracker uses the same five-hour-stretch definition referenced in this research and surfaces your baby's trajectory against AAP-aligned norms.
Explore the Wermom App