Background: Very few studies have investigated psychopathology in children of mothers with eating disorders (ED). We aimed to determine the effect of maternal ED on childhood psychopathology in a large population-based cohort and investigate relevant risk pathways using structural equation modeling (SEM). Methods: Data on emotional and behavioral problems at 31/2 years were obtained prospectively on 8,622 children from the Avon Longitudinal Study of Parents and Children (ALSPAC). Children of exposed women who self-reported lifetime anorexia nervosa (AN, N = 193) or bulimia nervosa (BN, N = 158) in pregnancy were compared with children of unexposed women (N = 8,271) using linear and logistic regression models. SEM was used to determine best-fitting risk models by child gender. Results: There was evidence that girls of AN women were more likely to have emotional, conduct, and hyperactivity disorders [Odds Ratio (OR): 1.7 (95% Confidence Intervals 1.0-3.0); OR: 2.2 (1.2-4.0); OR: 1.8 (1.1-3.1), respectively] and boys of AN women to have emotional disorders compared with unexposed [OR: 2.0(1.2-3.4)]. Girls of women with BN were more likely to show hyperactivity [OR: 1.7 (1.0-3.1)]; and boys to show emotional and conduct disorders compared with unexposed [OR: 2.2 (1.2-3.9); OR: 2.4 (1.4-4.2), respectively]. SEM models showed that pregnancy anxiety and depression mediated the effect of maternal ED on child psychopathology. Conclusions: Maternal ED are associated with different childhood psychopathology outcomes in boys and girls. Pregnancy anxiety and depression and active ED symptoms are important mediators of risk and are preventable; the direct effect of maternal lifetime ED was small.