Objective: To evaluate the long-term effects of joint protection on health status of people with early rheumatoid arthritis (RA).Design: A four-year follow-up of a randomized, controlled, assessor-blinded trial was conducted.Setting: Two rheumatology outpatient departments.Participants: People with rheumatoid arthritis less than five years since diagnosis.Interventions: Two 8-hour interventions were originally compared: a standard arthritis education programme, including 2½ hours of joint protection based on typical UK occupational therapy practice (plus 5½ hours on RA, exercise, pain management, diet and foot care); and a joint protection programme, using educational–behavioural training.Main measures: Adherence to joint protection, pain, hand pain on activity, Arthritis Impact Measurement Scales 2 and Arthritis Self-efficacy were recorded at 0 and 4 years.Results: Sixty-five people attended the joint protection and 62 the standard programmes. Groups at entry were similar in age (51 years; 49 years), disease duration (21 months: 17.5 months) and use of nonsteroidal anti-inflammatory and disease-modifying drugs. At four years, the joint protection group continued to have significantly better: joint protection adherence (p = 0.001); early morning stiffness (p = 0.01); AIMS2 activities of daily living (ADL) scores (p = 0.04) compared with the standard group. The joint protection group also had significantly fewer hand deformities: metacarpophalangeal (MCP) (p = 0.02) and wrist joints (p = 0.04).Conclusion: Attending an educational-behavioural joint protection programme significantly improves joint protection adherence and maintains functional ability long term. This approach is more effective than standard methods of training and should be more widely adopted.