A randomised treatment design for 49 chronically obsessive–compulsive ritualising patients was devised and three controlled comparisons were made. 1. During 7 weeks of self-exposure instructions, clomipramine treatment improved some measures of rituals and depression significantly more than did placebo medication; this effect was transient and disappeared as drug treatment and exposure were continued for a further 15 weeks. 2. During 11–16 weeks of clomipramine treatment, self-exposure instructions yielded highly significantly more patient improvement than did anti-exposure instructions on nearly all measures of rituals and some of social adjustment. 3. Adding therapist-aided exposure (1.3 hours) to self-exposure instructions (3 hours) after 8 weeks had a barely significant transient effect of dubious clinical value, which was lost by the end of exposure (at week 23) and during follow-up assessments to week 52. We conclude that of the three therapeutic factors tested, self-exposure was the most potent; clomipramine played a limited adjuvant role, and therapist-aided exposure a marginal one.