All topics › Development learning
How wide are the normal ranges for developmental milestones, and which delays genuinely warrant assessment?
How wide are the normal ranges for developmental milestones — and which delays genuinely warrant assessment, versus which are just slower-but-normal?
The normal ranges are much wider than milestone charts suggest. In the WHO's large multi-country study, healthy children walked alone anywhere from about 8 to nearly 18 months — and that whole span counted as normal [1]. The US revised its checklists in 2022 so each milestone sits at the age where at least 75% of children have reached it: a missed milestone now means "worth a conversation and possibly a screening", not "something is wrong" [2][3]. Early intervention helps modestly — the best trials (in preterm infants) show small-to-moderate gains that partly fade by school age [4]. Your own concerns about your child's development are worth taking seriously: properly elicited, they're about as accurate as formal screening tests [5].
The best single picture of normal variation comes from the WHO Multicentre Growth Reference Study, which followed 816 healthy children in Ghana, India, Norway, Oman, and the USA, with trained observers checking development monthly [1]. The researchers built "windows of achievement" for six gross motor milestones, bounded by the 1st and 99th percentiles — meaning the windows describe where essentially all healthy children fall, not where the average child falls:
| Milestone | Normal window (months) |
|---|---|
| Sitting without support | 3.8 – 9.2 |
| Standing with assistance | 4.8 – 11.4 |
| Hands-and-knees crawling | 5.2 – 13.5 |
| Walking with assistance | 5.9 – 13.7 |
| Standing alone | 6.9 – 16.9 |
| Walking alone | 8.2 – 17.6 |
Two details matter. First, the windows overlap: there is no single "correct" order that every child must follow. Second, 4.3% of the healthy children in the study never did hands-and-knees crawling at all [1] — skipping crawling is an observed normal variant, not on its own a sign of a problem.
The US Centers for Disease Control revised its milestone checklists in 2022 (published in Pediatrics), changing something fundamental: milestones moved from the average age (the 50th percentile — the age by which half of children have done it) to the age by which at least 75% of children have done it [2]. The reasoning is explicit: with a 50th-percentile milestone, a child missing it is still perfectly average, which feeds the "let's wait and see" reflex. With a 75th-percentile milestone, a miss is deliberately designed to be more actionable — a prompt to screen, not a diagnosis [3]. About a quarter of old milestones were dropped and 41% replaced; of the third of retained milestones that moved to different ages, two-thirds shifted older [2].
Two honest caveats. The social-emotional and cognitive milestones have the least normative data behind them — even the working group admitted the gaps [2]. And "actionable" does not mean "pathological": the CDC stresses these checklists support surveillance, not diagnosis [3].
The best evidence on intervening early comes from a 2024 Cochrane review of 25 randomized trials of developmental programmes for preterm infants after hospital discharge [4]. The headline: programmes probably improve cognitive and motor outcomes in infancy (low-certainty evidence), produce a solid cognitive gain by preschool age (about half a standard deviation on IQ-type tests, high certainty), but show probably no motor benefit by preschool age — and by school age, the evidence is too weak to say the gains last [4]. The infancy effects are small: roughly a third of a standard deviation for cognition, smaller for motor skills [4].
Two caveats worth knowing. These were preterm infants — high-risk babies, not the average child who missed one milestone. And blinding is impossible in these trials (parents know they're in a programme), so part of the effect may be the support and attention parents receive rather than the programme's exercises [4]. "Act early" is reasonable advice; "the earlier the better, always" is stronger than the evidence.
Frances Glascoe's validation work (971 children) found that parents' concerns about development, properly elicited, are about as accurate as quality screening tests — and that accuracy doesn't depend on the parent's education or experience [5]. This is not "trust your gut over the doctor": the finding is that systematically asked concerns (the kind a health visitor's questionnaire captures) predict problems well. A vague 3am worry and a specific observation ("she's not babbling at all at 14 months") are different things — but a specific observation deserves a professional's ear, not reassurance to wait.
The anxiety is the point, not a side effect. Milestone charts were designed to catch delays early, and they do create a second, unmeasured outcome: parental worry. Whether chart-watching itself raises anxiety has never been studied rigorously — a genuine gap in the evidence, given how many parents live inside it. If comparing your baby to the chart is making you miserable rather than informed, that is a legitimate reason to change how you use the chart, not a reason to feel guilty.
Disagreements between parents are normal here. One parent's "she'll get there" and the other's "we should call someone" are both reasonable stances given the evidence above — the ranges really are wide, and acting early is really worthwhile when something is off. The tiebreaker the evidence supports: when in doubt, ask the health visitor. That's what the surveillance system is for [6].
Care outside the scheduled reviews is possible. In the UK, developmental concerns go through your GP and health visitor, and you don't need to wait for the next routine review to raise one [6]. If you ever seek an assessment outside the NHS, take the reports back to your NHS team so the care joins up.
Fathers, sleep, and work barely appear in this evidence base. No study here measured what a developmental worry does to parental sleep, mental health, or employment. The honest summary: we know the system is designed around your observations [5], and we don't know what it costs you to make them.
Developmental surveillance doesn't have binary outcomes the way a drug trial does, so there are no "X in 1,000" figures here — the table below gives the closest absolute terms the evidence allows.
| Action or finding | What the evidence shows | Evidence rating |
|---|---|---|
| A child missing one milestone on a ≥75% checklist | By construction, about 250 in 1,000 typical children miss any single item (a mathematical consequence of the 75% cutoff, not an observed risk); it warrants a screening conversation, not a diagnosis [2][3] | C (expert consensus with normative data) |
| Early intervention (preterm infants) | Cognitive gain ≈ ½ SD by preschool age; infancy gains smaller; motor gains probably not sustained to preschool; school-age effects uncertain [4] | B (RCTs, low-to-high certainty by outcome) |
| Parent-raised concerns (systematically elicited) | Accuracy comparable to quality screening tests, across education levels [5] | B (validation studies, observational) |
| "Watch and wait" for a single slow milestone | No trial evidence of harm or safety; the WHO windows show wide normal variation [1] | D (no direct evidence) |
In the UK: your GP or health visitor — you can contact them between scheduled reviews [6]. Raise it promptly (don't wait for the next review) if: