Given the enormous influence of classification on the major clinical, research, and administrative activities of mental health professionals, understanding the true number and nature of disorders and the reasons for their comorbidity is an important public health priority. However, while studies of latent structure have yielded valuable information about disorder boundaries, their reliance on non-representative samples and failure to evaluate the practical implications of structural findings has limited their ability to effect nosological change. Conversely, community epidemiology studies, which inform classification by assessing the implications of diagnostic criteria in representative samples, have been limited by their focus on mental disorders as they are currently conceptualized by the field rather than on correlates and consequences of these disorders as they actually exist in nature. I consider the potential value of integrating systematically the methods of structural research with the methods of epidemiological research, exploring five ways in which these largely independent traditions may profitably be combined to inform the next classifications of mental disorders. By capitalizing on the complementary strengths of structural and epidemiological research, an integrated approach has significant potential to advance understanding of the nature of psychopathology and improve the validity and utility of its diagnosis.