Abstract Objective To identify the maternal body composition parameters that independently influence birth weight. Study design A longitudinal prospective observational study in a large university teaching hospital. One hundred and eighty-four non-diabetic caucasian women with a singleton pregnancy were studied. In early pregnancy maternal weight and height were measured digitally in a standardised way and the body mass index (BMI) was calculated. At 28 and 37 weeks’ gestation maternal body composition was assessed using segmental multifrequency bioelectrical impedance analysis. At delivery the baby was weighed and the clinical details were recorded. Results Of the women studied, 29.2% were overweight and 34.8% were obese. Birth weight did not correlate with maternal weight or BMI in early pregnancy. Birth weight correlated with gestational weight gain (GWG) before the third trimester ( r = 0.163, p = 0.027), but not with GWG in the third trimester. Birth weight correlated with maternal fat-free mass, and not fat mass at 28 and 37 weeks gestation. Birth weight did not correlate with increases in maternal fat and fat-free masses between 28 and 37 weeks. Conclusions Contrary to previous reports, we found that early pregnancy maternal BMI in a non-diabetic population does not influence birth weight. Interestingly, it was the GWG before the third trimester and not the GWG in the third trimester that influenced birth weight. Our findings have implications for the design of future intervention studies aimed at optimising gestational weight gain and birth weight. Condensation Maternal fat-free mass and gestational weight gain both influence birth weight.