This study uses information collected on two occasions from a probability sample of families with 8- to 12-year-old children (N= 718) participating in a general population study in 1983 and follow-up in 1987. It focuses on the association between maternal depressive symptoms and emotional disorder in children and adolescents, taking into consideration the influences of informant rating errors, contextual variables (economic disadvantage and family dysfunction), and child gender. Covariance structure analysis revealed a strong association between maternal depressive symptoms in girls (β= .59 in 1983 and β= .39 in 1987) but not in boys. This association is independent of the impact of contextual variables and the treatment of teacher rating errors. Among adolescent boys, maternal depressive symptoms are correlated with mother rating errors, suggesting the possible presence of maternal bias.