CONTEXT Obsessive-compulsive disorder (OCD) is clinically heterogeneous, but it is unclear whether this phenotypic heterogeneity reflects distinct, or partially distinct, etiologic mechanisms. OBJECTIVE To clarify the structure of the genetic and environmental risk factors for the major symptom dimensions of OCD. DESIGN Self-report questionnaires and multivariate twin model fitting. SETTING General community. PARTICIPANTS A total of 4355 female members of the TwinsUK adult twin register. MAIN OUTCOME MEASURES Scores on the Obsessive-Compulsive Inventory-Revised and 5 of its subscales (checking, hoarding, obsessing, ordering, and washing). RESULTS A common pathway model did not fit the data well, indicating that no single latent factor can explain the heterogeneity of OCD. The best-fit multivariate twin model was an independent pathway model, whereby both common and unique genetic and/or environmental factors contribute to the etiology of each symptom dimension. The hoarding dimension had the lowest loading on the common factor and was more influenced by specific genetic effects (54.5% specific). With the exception of hoarding, most of the genetic variance was due to shared genetic factors (ranging from 62.5% to 100%), whereas most of the nonshared environmental variance was due to dimension-specific factors. CONCLUSIONS Obsessive-compulsive disorder is unlikely to be an etiologically homogeneous condition. There is substantial etiologic overlap across the different OC symptom dimensions, but dimension-specific genetic, and particularly nonshared environmental, factors are at least as important. Hoarding shares the least amount of genetic liability with the remaining symptom dimensions. The results have implications for the current deliberations regarding OCD and the inclusion of a putative hoarding disorder in DSM-5.