An underlying assumption of self-care interventions is that they are most effective when reinforced. To test this assumption, 8 months after baseline, 589 subjects who had taken a 6-week Arthritis Self-management Course (ASMC) were randomized to (1) receive a bi-monthly arthritis newsletter, (2) attend a new 6-week Arthritis Reinforcement Course (ARC) or (3) receive no reinforcement. Between 8 and 20 months there were no significant differences among the three randomized groups. The results were unaltered by inclusion of assumed data of no change for the 46 subjects who did not complete the full 20-month study. Between baseline and 20 months all participants reduced their pain by 20%, depression by 14%, and visits to physicians by 35% (P less than 0.01). There were no trends toward loss of these effects over time. These findings indicate that the effects of a self-care intervention were sustained over 20 months and that the tested forms of reinforcement did not alter those effects.