OBJECTIVE: Optimal diagnostic thresholds were determined for DSM-IV attention deficit hyperactivity disorder, and the psychometric properties were compared to alternative definitions. METHOD: Structured diagnostic interviews of multiple informants for 380 clinic-referred youths aged 4-17 years were conducted. In addition, standardized clinicians' validation diagnoses of attention deficit disorder were obtained to assess agreement with clinical judgment. Measures of impairment were obtained to assess the accuracy of identifying youth with an impairing condition. RESULTS: Three subtypes of attention deficit hyperactivity disorder (predominantly inattentive, predominantly hyperactive-impulsive, and combined types) were distinguished on the basis of the degree of deviance on separate dimensions of inattention and hyperactivity-impulsivity. These three subtypes were found to differ in terms of types of impairment, age, and sex ratio, but not ethnicity. In terms of case identification of attention deficit hyperactivity disorder, DSM-IV was found to be very similar to DSM-III-R, except that DSM-IV identified more impaired girls and preschool children. CONCLUSIONS: These results support the decision to subdivide the heterogeneous category of DSM-III-R attention deficit hyperactivity disorder into three subtypes. The resulting DSM-IV definition appears to be somewhat less biased toward the symptom pattern typical of elementary school boys.