--- - |2+ Background Structural models of oppositional defiant disorder (ODD) symptoms have gained empirical support but await greater empirical scrutiny on issues such as reconciliation between competing models, evidence for gender differences, and delineation of external correlates. More extensive validation evidence is particularly necessary in consideration of their incorporation in psychiatric nomenclature. Methods We fitted previously proposed, but competing, models to ODD symptoms assessed with the Computerized Diagnostic Interview Schedule for Children (Shaffer et al., 2000, J. Am. Acad. Child Adolesc. Psychiatry, 39, 28) in a community sample and then fitted the best‐fitting model in a replication sample (combined N = 730, Mage = 9.89, SD = 0.75). Analyses also examined potential classes based on resulting subfactors, gender differences, longitudinal associations with later behavioral problems, and concurrent personality associations. Results Burke's (2010) two‐factor model composed of Irritability and Oppositionality subfactors best fit the data. Irritability and Oppositionality showed convergent and divergent patterns of association with personality traits at T1 and with externalizing and internalizing problems at T2. Latent class analyses revealed three classes (low severity, irritable/combined, and oppositional) which showed parallel divergence in externalizing and internalizing problem comorbidity. These findings were largely robust across gender and samples. Conclusions These findings support Irritability/Oppositionality subfactors of ODD in two mixed‐gender samples, demonstrate strong evidence for a lack of gender differences in such subfactors, and demonstrate their convergent and divergent validity in emergent latent classes, later behavioral problems, and personality correlates. - Journal of Child Psychology and Psychiatry, Page 18-29, October 2018.