Although it has been postulated that symptom subtypes are potential predictors of treatment response, few data exist on the longitudinal course of symptom and subtype categories in obsessive-compulsive disorder (OCD). Putative subtypes of OCD have gradually gained more recognition, but as yet there is no generally accepted subtype discrimination. Subtypes, it has been suggested, could perhaps be discriminated based on autogenous versus reactive obsessions stemming from different cognitive processes. In this study, our aim was to assess whether symptom and subtype categories change over time. Using the Yale-Brown Obsessive Compulsive Symptom Checklist (Y-BOCS-SC), we assessed 109 patients who met DSM-IV criteria for OCD to establish baseline values, then reassessed 91 (83%) of the initial group after 36 +/- 8.2 months. Upon reassessment, we found significant changes from baseline within aggressive, contamination, religious, symmetry and miscellaneous obsessions and within checking, washing, repeating, counting and ordering compulsion categories. Sexual, hoarding, and somatic obsessions, and hoarding and miscellaneous compulsions, did not change significantly. In accordance with the relevant literature, we also assigned patients to one of three subtypes-autogenous, reactive, or mixed groups. Though some changes in subtype categories were found, no subtype shifts (e.g., autogenous to reactive or reactive to autogenous) were observed during the course of the study. Significantly more patients in the autogenous group did not meet OCD criteria at follow-up than did patients in the other groups. Our results suggest that the discrimination between these two types of obsession might be highly valid, because autogenous and reactive obsessions are quite different, both in the development and maintenance of their cognitive mechanisms, and in their outcome.