Introduction We tested whether birth weight might be associated with gluteofemoral fat mass and insulin sensitivity later in life. Materials and methods Body size trajectory since birth, body composition at age 20, and markers of insulin resistance were measured in 316 Japanese women. A subset of 148 women underwent a 75 g oral glucose tolerance test. Multiple linear regression analyses were used to identify most important determinants of birth weight. Results Birth weight was correlated positively with height and weight at age 12, 15, and 20 years (all p < 0.001 except for weight at 12 years, p = 0.03). Although it showed no correlation with BMI at age 12 and 15, it was correlated positively with current BMI (p = 0.006). It showed positive correlations with lean mass in arms, legs, trunk, and the whole body at age 20 (all p < 0.001). Additionally, it was correlated positively with leg (gluteofemoral) fat mass (p = 0.007), although there was no correlation with total body and trunk fat mass. Furthermore, weight at birth showed inverse correlations with 2-h postglucose insulin concentrations (p = 0.008) whereas it was not correlated with fasting insulin and homeostasis model assessment-insulin resistance. In a multiple regression analysis, which included anthropometric and biochemical variables as independent variables, appendicular muscle mass (standardized beta 0.394, p < 0.001) emerged as a single determinant of birth weight (R-2 = 0.15). In a model which included gluteofemoral fat mass and 2-h postglucose insulin, birth weight was associated with gluteofemoral fat mass (standardized beta 0.240, p = 0.003) and 2-h postglucose insulin concentrations (standardized beta - 0.217, p = 0.007) (R-2 = 0.09). Conclusions Birth weight was associated positively with gluteofemoral fat mass and inversely with 2-h postglucose insulin concentrations, a marker of insulin resistance.