Objective: The authors studied factors associated with short-term treatment response in 38 nondepressed subjects with DSM-III-R obsessive-compulsive disorder (OCD). Method: The subjects completed 12 weeks of treatment with paroxetine (N = 20), placebo (N = 8), or cognitive-behavioral therapy (N = 10). Clinician and self-rated measures were gathered at baseline, during treatment, and after treatment. Results: Seventeen (45%) subjects had "much" or "very much" improvement and achieved at least a 40% decrease in their total Yale-Brown Obsessive Compulsive Scale (Y-BOCS) score. Responders had lower obsessive-compulsive scores on the Symptom Checklist 90-Revised, had a lower checking score on the Maudsley Obsessive-Compulsive Inventory, were less likely to have had prior drug therapy, and in general suffered more obsessive-compulsive symptoms. They were significantly less likely to have hoarding obsessions and corresponding compulsions. The latter finding was confirmed using multiple regression analysis. Conclusion: Hoarding is an important symptom that predicts poor treatment response in patients with OCD.