Women who are overweight or obese prepregnancy have shorter durations of producing milk (PM) and feeding breast milk exclusively (FBM-ex) than normal-weight women. We proposed that infant size at birth may reduce the negative associations between prepregnancy BMI and the durations of PM and FBM-ex. We used data from 2798 participants in the Infant Feeding Practices Study II and characterized infant size at birth as weight-for-gestational age (WGA). To assess possible mediation of the associations between maternal BMI and the durations of PM and FBM-ex by infant size at birth, Baron and Kenny's methods, the Sobel test, and bootstrapping were used. As expected, prepregnancy BMI was negatively associated (P < 0.0001) with the durations of PM and FBM-ex; it also was positively associated (P < 0.0001) with infant size at birth. However, infant WGA was positively associated (P < 0.0003) with the durations of PM and FBM-ex after adjustment for BMI. Thus, the negative associations between BMI and the durations of PM and FBM-ex were reduced by infant WGA; i.e. the statistical removal of infant size at birth increased the magnitude of the negative associations between BMI and the durations of PM and FBM-ex. Thus, the tendency of heavier mothers to deliver heavier infants reduces the true magnitude of the association between maternal prepregnancy BMI and shortened breastfeeding duration.