Tummy time has a number. Floor play doesn't.

The recommendation that came without a number

We make play mats, so we spend a lot of time thinking about what happens on floors. We went looking for evidence that babies aren't getting enough floor time. We did not find it — not because the floor time is there, but because almost nobody has ever measured it. What we found instead was a gap in the evidence itself, and we are publishing that instead of the story we expected.

The World Health Organization's advice for babies under one begins with floor-based play and adds tummy time as a qualification. Tummy time got the number — thirty minutes a day — and with it, all the measurement. Floor-based play got no number, and in the years since, it has been measured with a duration in exactly one published study.

Play on Rolly · August 2026 · Download the data: tummy time prevalence (CSV) · floor play measurement (CSV) · Zenodo archive (DOI)

On safe sleep. Tummy time means awake and supervised. Babies should always be placed on their backs to sleep. The Lullaby Trust's guidance on safer sleep is here, and nothing in this piece qualifies it.

---

What the guideline actually says

Here is the World Health Organization's 2019 recommendation for infants under one year old, in full and in its original order:

"Physically active several times a day in a variety of ways, particularly through interactive floor-based play; more is better. For those not yet mobile, this includes at least 30 minutes in prone position (tummy time) spread throughout the day while awake."

Read it as a sentence rather than as a slogan and the structure is plain. Floor-based play is the recommendation. Tummy time is a clause inside it, scoped to a particular group — infants who cannot yet move themselves — and introduced by the words this includes. Tummy time is an example of floor-based play, not a synonym for it, and it applies for a window of months rather than for the year.

One of those two things was given a number. The other was not.

It is worth being precise about what the number rests on. WHO classifies this as a strong recommendation on very low quality evidence, using the GRADE system. That is WHO's own assessment, published alongside the guideline, not our characterisation of it. Note also what the evidence review actually tested: not the thirty-minute threshold, but awake, supervised prone positioning as such. It reported favourable associations for gross motor development, total development, BMI-z and brachycephaly; an indeterminate result for plagiocephaly; and no association for fine motor skills or communication. All of those are observational associations, drawn mostly from studies that asked parents to remember. Throughout this piece, "tummy time" means awake, supervised prone positioning, and nothing else.

What happened next

Numbers are what health systems can act on. They can be put in a policy, printed on a leaflet, and — crucially — measured. An unquantified recommendation cannot be audited, so nobody audits it.

You can watch this happen. In 2025 an international panel ran a Delphi process to agree the indicators for an early-years movement report card. They went to the WHO guidelines to derive their list, and they came back with tummy time as an indicator. The phrase "floor-based play" does not appear anywhere in their paper.

The wider pattern is documented. Writing about Europe's failure to monitor the under-fives guidelines, the SUNRISE Study Europe group put it plainly: the lack of surveillance means that a high prevalence of not meeting the guidelines is, in effect, invisible, and invisibility leads to policy inaction. That is the general case. What follows is the specific one.

Six countries, six different questions

Tummy time has been measured, with a dose, in six countries: the United Kingdom, the United States, Australia, Canada, China and Singapore. That is the complete list, as far as we can establish. We searched for India, Brazil, Japan, Germany, the Nordic countries, Ireland, Spain and a dozen others and found nothing. We have not registered a formal search protocol, so treat this as a diligent search rather than a systematic review: if a seventh country exists we would rather be sent it than defend the number.

Here is what those six studies report, and — just as importantly — what each one actually asked.

In the two studies that asked about prone position alone and reported a typical dose, the median sits in single or low double digits — 8.6 to 15 minutes a day across the UK's first year, 10 to 12 in China. Where a published figure is higher, either the construct or the instrument is different.

That is a narrower sentence than the one we expected to write, and the narrowing is the point. It would be easy to say that most infants everywhere fall short of thirty minutes. Our own table will not support it: Canada reports 91.2% meeting the guideline at six months, Singapore 76.8%, and Italy over 85% practising it at all. Read those three as a league table and the shortfall vanishes. Read them carefully and they are not measuring the same thing as each other, let alone as the UK.

What the evidence does not support is the next step, which is being taken anyway.

The comparison already being made

In 2024, a study of Chinese infants reported that 27.2% met the guideline, and then wrote this:

"These findings are consistent with recent studies conducted in countries that have proposed guidelines for children's movement behaviour, which indicated that 31% of infants in the UK and 29.7% of infants in Australia achieved the recommended daily duration of tummy time."

Three countries, side by side, declared consistent. No mention of what each study asked, whom it asked, how old the babies were, or which year it was. We are not picking on that paper — it is a careful piece of work and its own limitations section is honest about its own sampling. We quote it because it is the moment the comparison entered the literature, and because the comparison does not hold.

Here is why.

One. The countries were asked different questions. Singapore's 76.8% — the highest of the figures that report a dose against a threshold — comes from asking about "tummy time or floor-based play". The UK study asked about the two separately, and reported tummy time at 8.6 to 15 minutes a day alongside floor play at 25.7 to 85.7 minutes a day. Apply Singapore's question to the UK's data and the UK figure would certainly rise — the UK's floor-play medians alone reach or exceed thirty minutes a day from four months. By how much cannot be calculated: the UK paper publishes the two distributions separately, never a combined per-child total. That the largest gap in the literature turns on a question nobody can resolve from the published tables is itself the point.

Two. The instrument decides the answer. Canada comes closest to a control experiment. One study asked the parents of 250 infants by questionnaire, and had a nested sub-sample of 94 keep a three-day time-use diary. At two months, 51.2% of the questionnaire sample met the guideline; 18.1% of the diary sub-sample did. At four months it was 68.8% against 40.4%; at six months, 91.2% against 62.8%. Across all three time-points together, the gap is starkest of all: 40.4% by questionnaire, 8.5% by diary — a difference of nearly five times. The study's own authors say the questionnaire tends to overestimate and that the diary is the measure suited to prevalence work. Every other country's figure is questionnaire-based.

Two honest qualifications. The diary group is a nested sub-sample rather than the same infants measured twice, and it is not demographically identical to the whole — 62.8% of it is Caucasian against 52.8% of the full sample — so some unknown share of the gap is composition rather than instrument. Even allowing for that, the gap between those two Canadian lines is larger than the entire distance between the UK, Australia and China.

Three. None of the six is a population. Not one of these studies sampled a country. The UK's figure rests on 109 infants recruited through social media during the COVID lockdowns, with age bands of 27, 48 and 34 children. China's participants were recruited through a commercial baby-class franchise — a channel one would expect to skew towards engaged, affluent families, though the study itself found no difference on this outcome between families attending its classes and those not — and it still came out at 27.2%. The US figure of 33.5% comes from a clinic population that was 86% Medicaid, and was measured at two months, the youngest of any study here. It also answers a different question: the study dichotomised at thirty minutes against AAP guidance, in data collected in 2010–2012, seven years before the WHO guideline existed.

Four. The ages do not match, and age drives the result. Adherence climbs steeply through the first year. In China, on the largest sample here at n=1040, it runs from 21.6% at 0–5 months to 32.5% at 6–11 months, a difference significant at p<0.0001. In Canada it runs 51.2% to 68.8% to 91.2% across two, four and six months. The US figure is measured at two months, Australia's at a mean of 3.6 months, and the UK's, China's and Singapore's are pooled across the whole first year. Ranking them is like ranking countries on reading while testing different school years.

Five. Fifteen years separate them. Australia's data was collected in 2008 and China's in 2023. The guideline itself only arrived in 2019, in the middle of the series.

And underneath all five, the recall windows differ. The studies did not ask about the same span of time. China asked what happened in the last twenty-four hours. The UK asked for time over an average week and divided it by seven, so its "minutes a day" is a derived weekly average rather than a day anyone lived. Singapore asked about a typical weekday or weekend. Canada's diary covered three days. These are different quantities before any of the five reasons above is applied — and the UK's weekly-average method is a further reason its figure cannot simply be set beside a twenty-four-hour recall.

Put numbers on it. The three studies the published comparison called consistent differ from each other by four percentage points. Within the single Canadian study, changing the instrument moves the figure by thirty-three points, and changing the age band moves it by forty. Changing the construct — Singapore's pooled question against the UK's prone-only one — spans forty-six, though that one is not held within a single study and so is an upper bound on the construct effect rather than an estimate of it. The methodological differences are an order of magnitude larger than the difference being attributed to countries.

Country	Figure	Unit	Age	n	Population	How it was measured	What was asked	Data year	Source
UK	31.2	%	0-11.9mo	109	convenience	parent-report online questionnaire (non-validated)	tummy time (prone only)	2020-2021	Hesketh KR & Janssen X, Int J Behav Nutr Phys Act 2022;19:58
China	27.2	%	0-11mo	1040	convenience	self-administered WeChat questionnaire, caregiver recall	tummy time (prone only)	2023	Huang Y et al., J Glob Health 2024;14:04048
Australia	29.7	%	mean 3.6mo	455	convenience	maternal report	tummy time (prone only)	2008	Hesketh KD et al., BMC Public Health 2017;17(Suppl 5):856
USA	33.5	%	2mo	859	clinic	parent report at well-child visit	tummy time (prone only)	2010-2012	Perrin EM et al., Pediatrics 2014;133(4):e857-67
Canada	51.2	%	2mo	250	regional cohort	parental questionnaire	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Canada	18.1	%	2mo	94	regional cohort	3-day time-use diary (validated against GENEActiv)	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Canada	68.8	%	4mo	250	regional cohort	parental questionnaire	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Canada	40.4	%	4mo	94	regional cohort	3-day time-use diary (validated against GENEActiv)	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Canada	91.2	%	6mo	250	regional cohort	parental questionnaire	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Canada	62.8	%	6mo	94	regional cohort	3-day time-use diary (validated against GENEActiv)	tummy time (prone only)	2018-2019	Carson V et al., Int J Behav Nutr Phys Act 2022;19:154
Singapore	76.8	%	0-11mo	219	convenience	caregiver report	tummy time OR floor-based play (POOLED)	2021-2022	Chua XH / Loo BKG et al., PLoS One 2024;19:e0298968
Italy	85.0	% (LOWER BOUND: source says over 85%)	infancy	3337	birth cohort	paediatrician-collected parental report	tummy time, BINARY PRACTICE (no dose)	2019-2020	Pandolfini C et al., Eur J Pediatr 2024;183:3029-38

What nobody counted

If tummy time has been measured in six countries, floor-based play — the recommendation it sits inside — has been measured, with a duration, in one.

Two studies asked about it and did not record how long: one American trial asked parents how often they played on the floor, and another recorded only whether any unrestrained floor time happened at all, finding that 34.4% of three-month-olds had any. Singapore folded it together with tummy time. A study that put accelerometers on 143 infants did not ask about it.

That leaves the UK's M-BAIT study, which measured both and reported both. Set its two series beside each other and the difference between the constructs is obvious.

UK, median minutes per day	0–3.9mo	4–7.9mo	8–11.9mo	12+mo
Infants in the band	27	48	34	58
Tummy time	8.6	15.0	10.0	0.0
Floor play	25.7	51.4	85.7	72.9

Those cell sizes matter, and we would rather print them than have them asked for. The 12-months-and-over band is also a separate sample from the other three: the 109 infants behind the prevalence figure are the under-twelve-month group, and the 58 in the final column are older children, aged twelve to eighteen months.

Tummy time falls to zero in that oldest band, which is consistent with the guideline scoping its clause to infants who are not yet mobile — and that band sits outside the recommendation's scope entirely, so a zero there is not evidence of compliance or of failure. WHO does not say prone play should stop; it stops specifying a dose. Floor play, meanwhile, carries on: it runs about three times larger than tummy time at the youngest two bands and roughly nine times larger by eight to twelve months. The recommendation with the longer life and the larger share of the day is the one that was never given a target.

Floor-based play has no threshold. No compliance rate can exist for it, and none is reported here or anywhere else. We are not claiming that children get too little floor play: the evidence cannot support that claim in either direction, and a company that sells play mats is the last party whose word anyone should take for it. This is not a finding about how much floor play children get. It is a finding that the question has not been asked.

Country	Study	n	Form of measure	Floor play measured?	Reported	Source
UK	M-BAIT	109	duration	yes, as a variable distinct from tummy time	floor play medians 25.7 / 51.4 / 85.7 / 72.9 min/day at 0-3.9 / 4-7.9 / 8-11.9 / 12+ mo; tummy time medians 8.6 / 15.0 / 10.0 / 0.0 min/day over the same bands	Hesketh KR & Janssen X, Int J Behav Nutr Phys Act 2022;19:58
Singapore	E-24 / IPRAMHO	219	pooled	pooled with tummy time, cannot be separated	76.8% met a combined "tummy time OR floor-based play" threshold	Chua XH / Loo BKG et al., PLoS One 2024;19:e0298968
USA	INSIGHT RCT	279	frequency	yes, but frequency only	asked "How often do you play on the floor with your baby?" at 8 and 20 weeks; no duration reported. ~95% of infants spent time in restrictive devices at both ages	Paul IM et al., Int J Behav Nutr Phys Act 2018;15:20
USA	Starting Early RCT	456	prevalence	partly, as prevalence not duration	82.6% ever practised tummy time; only 32.0% practised it on the floor; 34.4% had any unrestrained floor time	Gross RS et al., Obesity 2017;25:920-7
USA	Accelerometer study, 2 SE US counties	143	not measured	no - tummy time only	restrictive-device index 155.1 (SD 124.7) min/day at 6-7mo; bouncer 74.8, car seat 56.8, stroller 30.0, playpen 47.4 min/day	Shull ER et al., Child Obes 2022;18:466-75

What parents say gets in the way

There is a reason to be careful with a phrase like "fall short", and it is worth stating before the last section. Falling short here means falling short of an expert consensus rated very low quality by the body that issued it. It does not mean falling short of a clinical threshold, and it does not describe any individual child.

It also does not describe parents who cannot be bothered. One small Brazilian study asked around forty parents whose infants were getting almost no prone time why that was. Their answers, in order: 70.7% were afraid the baby would become breathless, 56.1% believed prone positioning was dangerous, 41.5% thought it was only safe once a baby could roll unaided, and 36.6% were afraid the baby would fall asleep face down. 14.6% said they were too busy.

We report those as findings from one small non-representative sample in one country, because that is what they are, and because this piece has spent two thousand words objecting to people generalising from exactly this kind of evidence. We are not going to explain why parents hold those beliefs; we do not know, and the study was not designed to tell us. What can be said is narrower and still useful: the reasons parents themselves give are about safety, not about time, and any advice that assumes a scheduling problem is answering a question they did not ask.

Awareness is thin on the specifics too. A quarter of US caregivers in one survey had not heard the prone-play recommendation at all, and in the Chinese study, 79.2% said tummy time was important while only 42.7% knew what the recommended amount was.

The number that isn't in the book

There is one place where this stops being an abstraction about measurement and becomes something a parent might notice.

The UK Chief Medical Officers' guidelines carry the number. For infants not yet mobile, they say, this includes at least 30 minutes of tummy time spread throughout the day while awake — and more is better. The NHS publishes it too, on its physical activity guidelines page for under-fives: "Try to include at least 30 minutes of tummy time spread throughout the day when they're awake."

So the number is public, and we want to be exact about that before saying anything else, because the interesting question is not whether Britain states the figure. It does. The question is where it turns up, and where it doesn't.

It does not turn up on Start for Life's "Baby moves" page — the NHS's own parent-facing guide to how a baby should move in the first year. That page says to do tummy time "every day, for a short amount of time", and to "gradually increase the amount of time you do this day by day". No duration appears on it.

And it does not appear in the Red Book — the personal child health record handed to parents, the book that follows a child through their early years. We text-searched the national-standard Personal Child Health Record, version 4.5, published December 2019, for the terms "tummy", "prone", "floor" and "physical activity". All four return zero occurrences. (Local trusts print variants, and we could not enumerate the content of the digital eRedbook, so this claim is about that edition and no other.)

The Red Book does contain a Safer Sleep page, and that guidance is correct and complete as safer-sleep guidance. Nothing here asks for it to change, and nothing here competes with it for space. The Red Book also contains a section called "Finding out about moving", which tracks motor milestones. There is a place in the record for how a baby moves. A tummy time prompt is not in it.

Two things should be said against reading this as a straightforward failure.

The first is that the omission may be deliberate and clinically sensible. "Baby moves" is a getting-started page aimed at newborns, and thirty minutes from birth would be the wrong advice — gradual build-up is what clinicians actually teach. The American Academy of Pediatrics does not use the WHO figure at all: it recommends 3 to 5 minutes, two or three times a day, working up to 15 to 30 minutes a day by around seven weeks. That is a different structure and a lower floor, from the largest paediatric body in the world. The thirty-minute figure is less of a global consensus than its repetition suggests — it is one WHO sentence, restated by the countries that adopted it, and not shared by the United States. It is also worth recalling that WHO rates it very low quality evidence.

Nor has the number travelled far into the places that would have to deliver it. A review of 43 early-years childcare policy documents across eight countries found tummy time mentioned in 36% of them — and of those, only 22% specified thirty minutes a day, while 48% said "at least daily" with no dose at all. An institution that declines to print a contested number is making a choice, not committing an oversight.

The second is that a parent-held record is a record, not a health-education leaflet, and the obvious answer is that tummy time is delivered through health visitor contacts instead. That answer is testable. The only study we found that tested it is American, and it looked at a US contact schedule rather than UK health visiting: 90% of mothers there received information about sleep positioning, against 55% postpartum and 26% at the two-month visit for awake positioning. That is a large asymmetry in the one system where somebody counted. It is not evidence about British health visitors, and we could find none. The objection stands untested here.

---

Authors. Compiled by Play on Rolly Ltd (company number 17144811). Named data steward: Anna Shevchuk (Play on Rolly Ltd). The dataset is deposited at Zenodo under DOI 10.5281/zenodo.22094615.

Funding. Self-funded by Play on Rolly Ltd. No external funding was received.

Competing interests. Play on Rolly Ltd manufactures and sells infant play mats and therefore has a direct commercial interest in the subject of this piece. Two specific conflicts should be named rather than left to inference. First, the company would benefit commercially from a finding that infants get too little floor play — a finding this piece does not make, because the evidence does not support it in either direction. Second, and less obviously, the company would benefit if floor-based play were quantified, monitored and added to parent-held records, which is the remedy the argument here implies. We think the argument stands on its own, but a reader is entitled to weigh it knowing who is making it.

Role of the funder. The funder is the author. Play on Rolly had full editorial control over this piece, including over which findings were reported and which were abandoned. No independent party held a veto. Readers should discount accordingly, and check the working, which is why it is published.

Sources. Every figure in this piece is transcribed from a published paper. We collected no new data.

• Prevalence figures: UK — Hesketh KR & Janssen X, Int J Behav Nutr Phys Act 2022;19:58. China — Huang Y et al., J Glob Health 2024;14:04048. Australia — Hesketh KD et al., BMC Public Health 2017;17(Suppl 5):856. USA — Perrin EM et al., Pediatrics 2014;133(4):e857–67. Canada — Carson V et al., Int J Behav Nutr Phys Act 2022;19:154. Singapore — Chua XH et al., PLoS One 2024;19:e0298968.
• Binary practice, not a prevalence-of-meeting figure: Italy — Pandolfini C et al., Eur J Pediatr 2024;183:3029–38.
• Floor play: the five studies in the accompanying CSV, including Shull ER et al., Child Obes 2022;18:466–75 for the container figures and Gross RS et al., Obesity 2017;25:920–7 for the unrestrained-floor-time figure.
• Evidence review: Hewitt L, Kerr E, Stanley RM, Okely AD, Tummy Time and Infant Health Outcomes: A Systematic Review, Pediatrics 2020;145(6):e20192168.
• Indicator selection: the 2025 Delphi consensus on early-years 24-hour movement report-card indicators, Sports Medicine – Open (PMC12331558).
• Surveillance: De Craemer M et al. for the SUNRISE Study Europe Group, Lancet Regional Health – Europe 2024 (PMC11129333).
• Childcare policy: Movement behavior policies in the early childhood education and care setting: an international scoping review, Front Public Health 2023;11:1077977.
• Barriers: e Silva et al., Int J Pediatrics 2023 — Brazil, around forty parents completing; treat as indicative only.
• Every row of both CSVs carries a DOI. All seven resolve; we checked them against Crossref rather than trusting our own transcription, having found two wrong.
• Guidelines: WHO, Guidelines on physical activity, sedentary behaviour and sleep for children under 5 years of age, 2019. UK Chief Medical Officers' Physical activity guidelines, 2019. American Academy of Pediatrics, Back to Sleep, Tummy to Play, healthychildren.org.
• NHS pages, all accessed 20 August 2026: nhs.uk/live-well/exercise/physical-activity-guidelines-children-under-five-years/ (states the 30-minute figure); nhs.uk/start-for-life/baby/baby-moves/ and its identical copy at nhs.uk/best-start-in-life/baby/baby-moves/ (state no duration).
• Personal Child Health Record: national-standard PCHR version 4.5, December 2019, RCPCH. Searched by text extraction of the full PDF for the terms "tummy", "prone", "floor" and "physical activity".
• Awareness: Zachry AH & Kitzmann KM, Am J Occup Ther 2011 (US); Huang Y et al., J Glob Health 2024 (China).
• Advice delivery: Koren A et al., J Pediatr Health Care 2010 — a US study of a US contact schedule, not UK health visiting.

What this is, and what it is not. This is a compilation. Its contribution is the assembly and the annotation, not new measurement. Every number was retyped from its source into a common schema; no source table, figure or passage is reproduced. We have added the columns the published comparison omitted — construct, instrument, population type, age band, sample size and year — and a comparable_group field recording which rows may legitimately be set beside one another.

Definitions. "Tummy time" is awake, supervised prone positioning. "Floor play" is unrestrained floor-based activity in any position. Where a study pooled the two, we say so and place it in its own comparability group.

Limitations. Every source is a convenience sample; not one is nationally representative. Parent report systematically overestimates tummy time relative to diaries and accelerometers, so the true measured picture is likely lower than the headline percentages rather than higher. Two figures that circulate elsewhere have been excluded because we could not verify a primary source for them: a pair of US four-month estimates, and a pair of Canadian medians. Our Red Book finding concerns the current national-standard edition of the personal child health record; local trusts print variants, and we could not enumerate the digital version's content, so we make no claim about it. The Canadian source reports its diary sub-sample as n=93 in its abstract and n=94 in its demographics table; we use 94 because it is the denominator that reproduces the published percentages. We have not published a formal search protocol for the claims that only six countries have dose-level measurement and that only one study reports floor-play duration; both rest on database and web searching that another worker could repeat but could not yet reproduce exactly, and we would treat a counter-example as a correction rather than a dispute. The evidence underlying the thirty-minute figure is rated very low quality by the body that issued it, which means "falling short of the guideline" should be read as falling short of an expert consensus, not of a clinical threshold.

Reproducing this. The dataset is published as two CSVs alongside this piece, with the compilation script and its tests, and archived at Zenodo under DOI 10.5281/zenodo.22094615 (CC0). Every row carries its citation and a link to the source. We would rather someone checked it than took our word for it — corrections to hello@playonrolly.co.uk.

---

Play on Rolly makes rollable play mats. That is why we were counting floor time, and we would rather say so than pretend this was disinterested. We think the conflict is worth stating twice: a company that sells mats has an obvious interest in a finding that babies need more floor play, which is precisely the finding the evidence would not support and which we are therefore not making. The compilation itself does not depend on us. Every figure comes from someone else's published work, the sources are named, and the working is public so that anyone can disagree with it properly.