Pregnant women who did the most physical activity were 30% less likely to gain more weight than recommended than those who did the least, according to a dose-response meta-analysis of 34 observational studies published on 11 October 2026 in the International Journal of Obesity. They were also 16% less likely to have a baby classed as large for gestational age. The relationship was roughly linear: more activity tracked with lower risk, with no threshold below which it stopped mattering. Because every included study was observational, the analysis shows an association, not that activity caused the difference.
What the analysis measured
The team, from Beijing Sport University, searched eight databases and included 34 independent studies of physical activity during pregnancy and two outcomes: excessive gestational weight gain, meaning gain above the recommended range for a woman’s pre-pregnancy weight, and delivering a large-for-gestational-age infant. Study quality was rated with the Newcastle-Ottawa Scale, and publication bias was tested with Begg’s and Egger’s tests.
The useful step is that the authors converted each study’s activity measure into one common unit, MET-hours per week, so studies using different questionnaires could be pooled on a single scale and modelled with restricted cubic splines. A MET-hour is one hour of activity at a given multiple of resting energy use. An activity rated at 4 METs, a value commonly assigned to brisk walking, done for 150 minutes a week is 4 × 2.5 = 10 MET-hours.
On that scale, women accumulating 10, 20 and 50 MET-hours a week had 4%, 9% and 20% lower risk of excessive weight gain than the least active. The association was strongest in the third trimester, where the same three levels corresponded to 8%, 16% and 35% lower risk. In the second trimester, no significant dose-response association was found.
All of these are relative reductions. The abstract does not give the baseline rate of excessive weight gain in the pooled studies, so a 20% relative reduction cannot be turned into a count of how many fewer women gained excess weight. The abstract also reports no confidence intervals for these estimates.
Why it matters, and what it changes
Pregnancy is one of the few stretches of adult life when weight-loss drugs are off the table. The FDA label for semaglutide as Wegovy, for example, says to stop it at least two months before a planned pregnancy. That leaves behaviour, including diet, activity and sleep, as what is left to study for weight during pregnancy, which is why a pooled analysis like this one gets attention.
The finding agrees in direction with earlier work linking activity in pregnancy to less excess weight gain. What this analysis adds is a shape: a straight-line relationship rather than a single cut-off, which suggests that differences in activity well below an athletic level were still associated with measurable differences in risk. It does not establish how much activity is safe or appropriate for any individual pregnancy, which remains a question for each woman and her clinician.
What is not yet known
Every study in the analysis was observational. Women who are more active in pregnancy may differ in many other ways, including diet, income and pre-pregnancy health, and pooling studies does not remove that confounding. Reverse causation is a specific risk for the third-trimester result: women who are already gaining more weight, or feeling unwell, may move less late in pregnancy, which would produce the same pattern without activity being the cause. Activity was largely self-reported, and harmonising different questionnaires into MET-hours adds measurement error. We could only read the abstract; the full paper, with its confidence intervals and the publication-bias results, was not reachable.
Sources: Lin Z, Zhang L, Li H, et al. “Physical activity during pregnancy and the risks of excessive gestational weight gain and large-for-gestational-age: a dose-response meta-analysis of observational studies.” International Journal of Obesity, published online 11 October 2026. doi:10.1038/s41366-026-02236-7 (PubMed 42860433). The authors declared no competing interests. More research coverage in News.
