Behavioral and health status outcomes of an unreinforced, self-help education program for arthritis patients taught by lay persons were examined in 2 ways: a 4-month randomized experiment and a 20-month longitudinal study. At 4 months, experimental subjects significantly exceeded control subjects in knowledge, recommended behaviors, and in lessened pain. These changes remained significant at 20 months. The course was inexpensive and well-accepted by patients, physicians, and other health professionals.