Education of patients with arthritis began with an emphasis on conveying knowledge, grew to include behaviour change, compliance, and more general coping and management of disease and then progressed to consider physical and psychosocial health outcomes. Research continues in all these areas. Control, in many forms (locus of control, self perceived efficacy, learned helplessness), is now suggested to be a central mediating variable. Evaluation of programmes is moving away from programme v usual care towards comparison of alternative methods of delivery and matching of method to learner. The first generation of researchers in arthritis education tended to be care givers with little formal education in behavioural sciences and evaluation methodology; the programmes they designed were often empirically based. The current generation, nurtured in large part by funds from the Arthritis Foundation and the National Institutes of Health, is better trained in designing programmes grounded in behavioural sciences and educational theory. In the long run, collaborations with care givers and patients will considerably strengthen the effectiveness of education programmes for patients. A variety of educational strategies have been shown to change the knowledge, behaviour, and health of patients with arthritis for the better. Many methods seem to work, so long as the programme is planned, has a goal, and is accountable. There is much work still to be done to teach care givers to be better teachers, and patients to be better managers of their diseases, in concert with their doctors, and to focus on high risk groups. Although most work has been done with patients with rheumatoid arthritis and osteoarthritis, many of these findings can and should be safely generalised to less studied rheumatic diseases. Finally, we need to consider the patient first as a person, and to provide education through all avenues, not just the medical care system.