Background: Maternal eating disorders (ED) have been shown to increase the risk of feeding difficulties in the offspring. Very few studies, however, have investigated whether the effect of a maternal ED on childhood feeding is a direct effect or whether it can be ascribed to other child or maternal factors. We aimed to determine the role of maternal anxiety and depression in mediating the risk for feeding difficulties in infants of women with ED. Methods: A prospective study comparing women with lifetime ED (441) and without any lifetime psychiatric disorder (10,461) and their infants from the Avon Longitudinal Study of Parents and Children (ALSPAC). We investigated the effect of: maternal anxiety and depression in late pregnancy (32 weeks) and the post-partum (8 weeks), child temperament and developmental status on infant feeding difficulties at 1 and 6 months. We also investigated the effect of active pregnancy ED symptoms. We tested 3 models and their fit to the data using structured equation modelling: a direct effect model, a fully mediational model and an integrated (partial mediational) model. Results: The integrated model including a direct effect of maternal lifetime ED on infant feeding and a mediational path via maternal distress (a latent variable combining anxiety and depression) fitted the data best. This also applied to maternal pregnancy ED symptoms. Feeding difficulties in turn increased maternal distress over time. Conclusions: Lifetime ED and active pregnancy ED increase the risk for infant feeding difficulties and do so via maternal distress (i.e., depression and anxiety). This has important implications for prevention and early intervention in relation to infant feeding difficulties, as well as for future research in the field.