Objective: To examine the influence of psychiatric comorbidity on response and relapse rates in children and adolescents treated with paroxetine for obsessive-compulsive disorder (OCD). Methods: Patients responding following 16 weeks of treatment (phase I) were randomized to continued paroxetine or to placebo for 16 additional weeks (phase II). OCD response (phase I) and relapse (phase II) criteria were based on the Clinical Global Impression-Improvement Scale and the Children's Yale-Brown Obsessive Compulsive Scale. The presence of OCD and other psychiatric disorders was ascertained using the Kiddie Schedule for Affective Disorders and Schizophrenia for School-Age Children-Present and Lifetime Version interview. Results: At entry, 193 of 335 (57.6%) patients had at least one psychiatric disorder in addition to OCD, and 102 of 335 (30.4%) had multiple other disorders. Although the response rate to paroxetine in the overall population was high (71%), the response rates in patients with comorbid attention deficit hyperactivity disorder, tic disorder, or oppositional defiant disorder (56%, 53%, and 39%, respectively) were significantly less than in patients with OCD only (75%) (intent-to-treat population, last observation carried forward analysis, p < 0.05). Psychiatric comorbidity was associated with a greater rate of relapse in the total patient population (46% for one or more comorbid disorders [p = 0.04] and 56% for two or more comorbid disorders [p < 0.05] vs. 32% for no comorbidity). Conclusions: The results of these post hoc analyses show that comorbid illness adversely impacted response to pharmacotherapy with paroxetine in pediatric OCD and significantly increased risk of relapse following withdrawal from treatment. Continued paroxetine treatment reduced the relapse rates in all groups compared with placebo, including those with comorbid illness. Because pediatric OCD is frequently comorbid with other psychiatric disorders, results of randomized, controlled pediatric OCD trials that use multiple exclusion criteria may not generalize to more naturalistic OCD samples.