OBJECTIVES: There is little information regarding motor development of children born at term with low birth weight (LBW), a group that constitutes a large proportion of children in South Asia. We used data from infancy and at school age from a LBW cohort to investigate children's motor performance using causal inference. DESIGN: Cross-sectional follow-up study. SETTING: Delhi, India. PARTICIPANTS: We recruited 912 children aged 5 years who had participated in a trial of vitamin D for term LBW infants in the first 6 months of life. OUTCOME MEASURES: We focused on gross motor development, using the Ages and Stages Questionnaire (ASQ) gross motor scale and several measures of motor performance. We examined the effects on these of current anthropometry, vitamin D status and bone health, controlling for age, sex, season of interview, socioeconomic variables, early growth, recent morbidity, sun exposure and animal food intake. RESULTS: In adjusted analyses, stunted children (height-for-age Z (HAZ) <-2) took longer to run 20 m (0.52 s, 95% CI 0.35 to 0.70; p<0.001) and had greater odds of a failing score on the ASQ (OR 3.00, 95% CI 1.41 to 6.38, p=0.004). Greater arm muscle area was associated with faster run time, and the ability to perform more stands and squats in 15 s. Poorer vitamin D status was associated with the ability to perform more stands and squats. Lower tibia ultrasound Z score was associated with greater hand grip strength. Early growth and current body mass index had no associations with motor outcomes. CONCLUSIONS: Current HAZ and arm muscle area showed the strongest associations with gross motor outcomes, likely due to a combination of simple physics and factors associated with stunting. The counterintuitive inverse associations of tibia health and vitamin D status with outcomes may require further research.