To examine symptom change over time, the effect of attrition on treatment outcome, and the putative mediators of cognitive therapy (CT) versus behavior therapy (BT) for obsessive-compulsive disorder (OCD) using archival data. Sixty-two adults with OCD were randomized to 20 sessions of CT (N = 30) or BT (N = 32) that consisted of 4 weeks of intensive treatment (16 hr total) and 12 weeks of maintenance sessions (4 hr). Independent evaluators assessed OCD severity using the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) at baseline and at Weeks 4, 16 (posttreatment), 26, and 52 (follow-up). Behavioral avoidance, depressive symptoms, and dysfunctional beliefs regarding responsibility were also measured at each assessment. Study hypotheses were tested using multilevel modeling. The slope of change in Y-BOCS scores was significantly greater in BT than in CT (d = 0.69), and those receiving BT had lower Y-BOCS scores at the final assessment than those receiving CT (d = 1.17). The greater slope of change in BT versus CT did not differ for dropouts versus completers. Reduction in depressed mood mediated changes in Y-BOCS across the 2 treatments, but a reduction in sense of responsibility and a decrease in avoidance did not. Instead, Y-BOCS improvements appeared to precede a decrease in avoidance. BT may have some therapeutic advantage over CT in the treatment of OCD, and this advantage does not appear to be due to a differential pattern of responding for treatment dropouts versus completers. Further, inconsistent with hypotheses, improvements in OCD symptoms were mediated by reductions in depressed mood instead of decreases in avoidance and responsibility. Theoretical, methodological, and clinical implications are discussed.