To examine the effectiveness and cost-effectiveness of symptomatic versus aggressive treatment in patients with established, stable rheumatoid arthritis (RA). A randomised observer-blinded controlled trial and economic evaluation with an initial assessment at randomisation and follow-ups at 12, 24 and 36 months. Five rheumatology centres in England. The 'symptomatic care' patients were managed predominantly in primary care with regular visits by a rheumatology specialist nurse. The 'aggressive care' patients were managed predominantly in the hospital setting. Patients with RA for more than 5 years were screened in rheumatology clinics. The symptomatic care patients were seen at home every 4 months by a rheumatology specialist nurse and annually by the rheumatologist. The aim of treatment was symptom control. The aggressive care patients were seen at least every 4 months in hospital. Their treatment was altered (following predefined algorithms) with the aim of suppressing both clinical and laboratory evidence of joint inflammation. The main outcome measure was the Health Assessment Questionnaire (HAQ). Others included the patient and physician global assessment, pain, tender and swollen joint counts, the erythrocyte sedimentation rate and the OSRA (Overall Status in Rheumatoid Arthritis) score. X-rays of the hands and feet were performed at the beginning and end of the study. The EQ-5D was used in the health economic evaluation. Comprehensive costs were also estimated and were combined with measures of outcome to examine between-group differences. A total of 466 patients were recruited; 399 patients completed the 3 years of follow-up. There was a significant deterioration in physical function (HAQ) in both arms. There was no significant difference between the groups for any of the clinical outcome measures except the physician global assessment [adjusted mean difference 3.76 (95% CI 0.03 to 7.52)] and the OSRA disease activity component [adjusted mean difference 0.41 (95% CI 0.01 to 0.71)], both in favour of the aggressive arm. During the trial, second-line drug treatment was changed in 77.1% of the aggressive arm and 59.0% of the symptomatic arm. There were instances when the rheumatologist should have changed treatment but did not do so, usually because of mild disease activity. The symptomatic arm was associated with higher costs and higher quality-adjusted life-years (QALYs). There was a net cost of 1517 Pounds Sterling per QALY gained for the symptomatic arm. Overall, the primary economic analysis and sensitivity analyses of the cost and QALY data indicate that symptomatic treatment is likely to be more cost-effective than aggressive treatment in 58-90% of cases. This trial showed no benefit of aggressive treatment in patients with stable established RA. However, it was difficult to persuade the rheumatologist and/or the patient to change treatment if the evidence of disease activity was minimal. Patients in the symptomatic arm were able to initiate changes of therapy when their symptoms deteriorated, without frequent hospital assessment. Approximately one-third of current clinic attenders with stable RA could be managed in a shared care setting with annual review by a rheumatologist and regular contact with a rheumatologist nurse. Further research is needed into disease progression and the use of biological agents, minimum disease activity level below which disease progression does not occur, cost-effectiveness through shared care modelling, the development of a robust and fail-safe system of primary-care based disease-modifying anti-rheumatic drug (DMARD) monitoring, and predicting response to DMARDs.