The textbook answer is "three months." Our cohort says the real median is closer to fourteen weeks — and three distinct resolution curves explain why some families feel relief at ten weeks while others wait until twenty.
The clinical definition of infantile colic still leans on the Wessel "rule of threes": crying more than three hours a day, more than three days a week, for more than three weeks, in an otherwise healthy and well-fed infant1. That definition, written in 1954, has held up clinically, but the implicit timeline that grew around it — colic peaks at six weeks, resolves by three months — is a simplification that does not fit the spread of real-world experience.
Recent literature has pushed back. A 2017 meta-analysis pooling 8,690 infants across 28 studies found prevalence falling sharply between 6 weeks and 12 weeks, but residual elevated crying past 12 weeks in roughly one in five infants — particularly in formula-fed cohorts2. The American Academy of Pediatrics' parent-facing guidance at HealthyChildren.org now describes colic as typically peaking "around 6 weeks of age" and "decreasing significantly after 3 to 4 months," language that quietly widened the window3.
Our research question: among parents actively tracking infant crying with the Wermom App, at what week does daily fuss duration drop below the Wessel threshold and stay below it? And do clinically distinguishable sub-patterns exist within the resolution curve?
Data source. Anonymized aggregated cry- and fuss-duration logs from Wermom App users with infants born between June 2024 and December 2025. Logs were de-identified at ingestion; only week-of-age, daily fuss minutes, feeding type, and resolution timestamps were retained for analysis.
Inclusion criteria. Infants with (a) at least 21 logged days between weeks 3 and 20, (b) at least one window meeting the Wessel rule of threes between weeks 3 and 10, and (c) continuous tracking through the resolution event. After filtering, the analytic cohort was n = 8,247 infants.
Definition of resolution. The first week in which fuss minutes fell below 60 per day on at least 5 of 7 days, with no Wessel-threshold rebound in the following 14 days. This is a more conservative bar than "the day crying dropped" — we wanted resolution that stuck.
Analysis. Median, IQR, and 90th-percentile resolution week were computed for the whole cohort and by feeding type. Sub-patterns were identified by k-means clustering (k=3) on smoothed weekly fuss-duration curves between weeks 3 and 20, with k selected via elbow plot. Sensitivity analyses re-ran with k=2 and k=4 to confirm the three-cluster solution was stable.
What we did not do. We did not collect physician-confirmed colic diagnoses; the cohort is parent-identified colic that meets the Wessel threshold in our logs. We did not collect outcomes beyond cry resolution (e.g., long-term sleep, attachment, parental mental health). We did not randomize anything — this is observational.
The popular "by three months it's gone" framing fits roughly the lower half of our cohort. Half of infants meeting the Wessel threshold had not yet hit our resolution definition at the 12-week mark, and one in ten was still above threshold past 18 weeks. For a parent reading "colic ends at three months" on a hospital discharge sheet, those extra 6–7 weeks can feel like a long, isolating tail.
Anchoring expectations matters clinically. Wermom IBCLC advisors flagged the same pattern they see in lactation visits: parents told "you have two weeks left" who then experience another month of evening fussing report higher rates of feeding-method second-guessing, formula switching, and unnecessary elimination diets. The 2017 meta-analysis by Wolke and colleagues warned of the same risk — that overly optimistic timelines drive intervention escalation when patience would have been sufficient2.
The NICHD's parent guidance, by contrast, is explicit that "babies cry for many reasons, and crying usually peaks at 6 to 8 weeks of age and then slowly decreases" — language that aligns more closely with our data and with the period-of-purple-crying framing endorsed by the National Center on Shaken Baby Syndrome4.
Clustering revealed three stable groups, each with a different practical implication.
Cry duration peaked at weeks 5–6 and fell below threshold by a median of week 10.8. These infants had the shortest peak intensity (mean peak 162 fuss-minutes/day) and the cleanest decline. Breastfed-only infants were over-represented in this cluster (61% vs 49% in the full cohort).
Cry duration peaked between weeks 6 and 8 and fell below threshold by a median of week 13.4. Highest peak intensity (mean peak 198 fuss-minutes/day). This is the cluster the textbook describes, and it is the largest group — but it is barely half of infants meeting the Wessel definition.
Cry duration peaked later (median peak week 9) and resolved by a median of week 17.6, with 18% still above threshold at week 20. Feeding-related co-symptoms — frequent spit-up, log-noted arching, formula switches — were 2.1× more common in this cluster than in Patterns A and B. This is the cluster pediatricians should consider when colic is still front-and-center past 12 weeks: it overlaps clinically with the GERD and cow's-milk-protein-intolerance evaluations described in AAP and NASPGHAN guidance5.
Parents living through Pattern C often report that their pediatrician told them "should be over by now" — and our data suggests "by now" simply moved. For deeper context on distinguishing reflux from GERD in this period, see Wermom's parent guide on reflux vs GERD in infants.
Median resolution week by primary feeding method at week 6:
Formula-fed infants in our cohort resolved roughly a week later on average and were over-represented in Pattern C, consistent with the 2017 meta-analysis2. This is observational — we did not measure cow's-milk-protein intolerance prevalence directly, and feeding-method selection is not random. Interpret as association, not cause.
We also stratified by maternal-reported sleep environment (room-sharing vs separate room) and by daily wake-window structure. Neither produced statistically meaningful resolution-week differences once feeding type was held constant.
For pediatricians. Consider quoting a wider resolution window at the 6-week well-child visit. "Most infants are noticeably better between 12 and 16 weeks, with about one in five still working through it past 16" maps to our data and matches recent peer-reviewed prevalence work. Re-evaluate at week 12 if crying remains above threshold rather than waiting for an unhappy week-16 visit.
For Pattern C families. If your infant is past 12 weeks and still meeting the Wessel threshold, and especially if you are noting feeding-related co-symptoms, a pediatric reassessment is reasonable. AAP guidance for persistent crying past 3 months explicitly lists GERD, cow's-milk-protein intolerance, and infant sleep-onset issues as differentials worth exploring3.
For all families. The most consistent evidence-backed intervention in colic remains parental support and crying-pattern education, not formula switching or supplements. The NIH NICHD's "Crying and Your Baby" resource is a clear, free, parent-friendly starting point4. For families tracking patterns at home, the milestone and feeding logs in the Wermom App let you visualize the fuss-duration curve in the same way our research team did, which can re-frame "is this ever going to end" into "we are 71% of the way down from peak."
Selection. Wermom App users skew toward data-engaged, English-speaking, US- and Canada-based parents with smartphone access. We cannot generalize to populations without app access, nor to non-English-speaking immigrant families where colic prevalence has been reported as different.
Measurement. Parent-logged fuss duration is subject to recall and definition variance. We mitigated this with same-day in-app logging and a conservative resolution definition, but a parent's "fussing" is not always the same as a researcher's.
No control group. Comparing to "infants who never met the Wessel threshold" would have required a different study design. The resolution curves we describe are within-colic, not vs non-colic.
No clinical diagnosis adjudication. We cannot rule out that some Pattern C infants had unrecognized GERD, food protein-induced allergic proctocolitis, or another organic driver. That is precisely why Pattern C is the cluster we recommend pediatricians take an extra look at.
Two follow-on studies are planned for the Wermom Research 2026–2027 calendar. First, a feeding-stratified analysis with finer-grained intolerance markers (stool-blood log incidents, formula-switch reasons). Second, a parent mental-health overlay using the same anonymized cohort to ask whether earlier expectation-setting (Pattern-based counseling at 6 weeks) is associated with lower PHQ-9 scores at 12 weeks. Both will be published openly here at wermomhealth.com/research with full methodology.
For parents reading this in the middle of a colicky stretch: you are not failing. The textbook line drew a finish line too close. Most of you are about a week and a half further out than you were told. You are still on the curve.
Dataset citation: Wermom Research. Anonymized infant cry-pattern aggregate, n=8,247 (2024-06 to 2025-12). Wermom Health, 2026. Available on request to research@wermomhealth.com.
Conflict of interest: The dataset is generated by users of the Wermom App, a product of Wermom Inc. The analysis was performed by Wermom's research team. No external funding.
The Wermom App's fuss log lets you plot your infant's daily cry minutes against the same week-of-age timeline used in this study.
Open Wermom App →