Objective: Comorbidity poses a major challenge to conventional methods of diagnostic classification. Although dimensional models of psychopathology have shed some light on this issue, the reason for interrelationships among dimensions is unclear. The current study employed an alternative approach to characterizing patterns of comorbidity among common mental disorders by modeling them instead as clusters by using latent class analysis (LCA). Method: Latent class analyses of Diagnostic and Statistical Manual of Mental Disorders diagnoses from two nationally representative epidemiological samples-the National Comorbidity Survey and National Comorbidity Survey-Replication datasets-were undertaken. Results: Within each dataset, LCA yielded 5 latent classes exhibiting distinctive profiles of diagnostic comorbidity: a fear class (all phobias and panic disorder), a distress class (depression, generalized anxiety disorder, dysthymia), an externalizing class (alcohol and drug dependence, conduct disorder), a multimorbid class (highly elevated rates of all disorders), and a few-disorders class (very low probability of all disorders). Whereas some disorders were relatively specific to certain classes, others (major depression, posttraumatic stress disorder, social phobia) appeared to be evident across all classes. Profiles for the five classes were highly similar across the two samples. When bipolar I disorder was added to the LCA models, in both samples, it occurred almost exclusively in the multimorbid class. Conclusions: Comorbidity among mental disorders in the general population appears to occur in a finite number of distinct patterns. This finding has important implications for efforts to refine existing diagnostic classification schemes, as well as for research directed at elucidating the etiology of mental disorders.