Previous epidemiological studies of correlates of child and adolescent mental disorders in the general population have focused more on child/adolescent and socioeconomic/sociodemographic characteristics than on family characteristics. Moreover, there are no generally accepted methods to analyze and interpret correlates. The purpose of the Quebec Child Mental Health Survey in this regard was twofold: (1) to identify correlates of DSM-III-R internalizing and externalizing disorders according to informant (youth, parent, teacher), for three age groups (6-8, 9-11, and 12-14 years), including relevant family characteristics not considered in previous studies; and (2) to interpret the relative importance of risk indicators by ranking correlates according to strength and consistency of association across age groups. Logistic regression models suggest the inconsistency of correlates across informants. The ranking of correlates reveals that individual and family characteristics make a more important contribution than do socioeconomic characteristics, thereby supporting the relevance of proximal variables in the development of psychopathology.