Predictors of improvement in obsessive-compulsive symptoms (Y-BOCS) in a randomized clinical trial with adult obsessive-compulsive disorder outpatients were examined. Results of multiple regression analyses revealed that a positive helping alliance was significantly predictive of posttreatment Y-BOCS. Treatment expectancy and high motivation to change were not significantly related to posttreatment outcome. None of the predictors were significantly related to Y-BOCS levels at 12-month follow-up, but positive alliance showed a trend to significance.