Evidence on whether nutritional supplementation affects physical activity (PA) during early childhood is limited. We examined the long-term effects of lipid-based nutrient supplements (LNS) on total PA, moderate-to-vigorous PA (MVPA) and sedentary behaviour (SB) of children at 4–6 years using an accelerometer for 1 week. Their mothers were enrolled in the International Lipid-based Nutrient Supplement-DYAD randomised controlled trial in Ghana, assigned to daily LNS or multiple micronutrients (MMN) during pregnancy through 6 months postpartum or Fe and folic acid (IFA) during pregnancy and placebo for 6 months postpartum. From 6 to 18 months, children in the LNS group received LNS; the other two groups received no supplements. Analysis was done with intention to treat comparing two groups: LNS v. non-LNS (MMN+ IFA). Of the sub-sample of 375 children fitted with accelerometers, 353 provided sufficient data. Median vector magnitude (VM) count was 1374 (interquartile range (IQR) 309), and percentages of time in MVPA and SB were 4·8 (IQR 2) and 31 (IQR 8) %, respectively. The LNS group (n 129) had lower VM (difference in mean −73 (95 % CI −20, −126), P = 0·007) and spent more time in SB (LNS v. non-LNS: 32·3 v. 30·5 %, P = 0·020) than the non-LNS group (n 224) but did not differ in MVPA (4·4 v. 4·7 %, P = 0·198). Contrary to expectations, provision of LNS in early life slightly reduced the total PA and increased the time in SB but did not affect time in MVPA. Given reduced social-emotional difficulties in the LNS group previously reported, including hyperactivity, one possible explanation is less restless movement in the LNS group.