We investigated the comparative efficacy of citalopram vs. citalopram administered with clomipramine, in treatment-resistant obsessive-compulsive disorder (OCD). Sixteen adult outpatients participated in a 90-day, randomized, open-label trial. Eligible patients were aged 18 to 45 years, had moderate to severe DSM-III-R OCD of >/= one year's duration, a baseline Yale-Brown scale (Y-BOCS) score of >/= 25 and no other active axis I diagnosis, and had failed adequate clomipramine and fluoxetine trials. The citalopram-plus-clomipramine group (n = 9) experienced a significantly larger percent decrease in mean Y-BOCS score by day 90 than the citalopram alone group (n = 7). Only one citalopram patient decreased her score by >/= 35%, and two by >/= 25%. All nine citalopram-plus-clomipramine patients experienced decreases of 35%. Side effects were mild to moderate in both groups. We also treated with citalopram six OCD patients who had not tolerated fluoxetine alone and clomipramine alone; three achieved Y-BOCS score decreases of >/= 35% at 90 days. Since citalopram does not significantly affect clomipramine metabolism, the improvement in the combined drug group is unlikely to have resulted from increased plasma clomipramine levels. Double-blind controlled trials are needed of citalopram in OCD, and of combining citalopram with clomipramine in treatment-resistant OCD.