The paper presents four alternative confirmatory factor models to represent the structure of measures of attention deficit and conduct disorder obtained from multiple sources. These models range in complexity from a simple one-factor model with no method effects to a two-factor model with method effects. These models were fitted to data gathered on maternal, teacher and self-report of attention deficit and conduct disorder for a sample of New Zealand children studied at ages 9 and 10 years. The analysis strongly suggests that the best representation of the data was a two-factor model in which there were distinct factors of attention deficit and conduct disorder with these factors being highly correlated (r = 0.87-0.88). It is suggested that the most suitable clinical nomenclature may be to describe conduct disorder and attention deficit generically as externalizing disorders, subscripting this classification with clauses which describe the relative contributions of conduct disorder and attention deficit. The implications of this finding for the measurement, classification and explanation of conduct disorder and attention deficit disorder are discussed.