OBJECTIVE: Evaluation of a complex and variable disease such as rheumatoid arthritis (RA) poses a challenge particularly over the medium to long term. A practical framework to evaluate clinically relevant outcomes over the long term is the "5D" approach of Fries, described in 1980. We describe the 20 year outcome in 52 survivors of a 123 patient cohort in terms of change in discomfort, disability, drug side effects, dollar costs, and deaths. METHODS: We studied 123 patients with RA allocated to their first disease modifying antirheumatic drug (DMARD) between 1977 and 1979. All were under the overall care of one physician over the 20 years and were maintained where possible taking a single DMARD. Baseline demographic variables, the Ritchie Articular Index (RAI), Lee functional index, and erythrocyte sedimentation rate (ESR) were initially recorded. The extent to which the demographic and disease variables contributed to need for joint replacement surgery was assessed. Therapies for comorbidity were also documented. RESULTS: At cohort inception mean age was 50 years, RAI was 35, and median disease duration 5.5 years. F:M ratio was 90:33; 96% of patients were positive for rheumatoid factor (RF). Initial median ESR was 55 mm/h. At 20 years, 9 patients (7% of original cohort, 14% of survivors) were lost to followup and 62 (50%) had died. In the 52 survivors RAI, a surrogate for disability, showed a significant improvement (p < 0.0001), but disability measured by Lee functional index showed a deterioration (p = 0.018); 50% underwent joint replacement surgery. Initial ESR and mean ESR over the first 10 years of followup were significantly higher in those who required surgery. Nonsteroidal antiinflammatory drug (NSAID) use declined, but at least 2 deaths and 4 renal deaths that may have been related to therapy were attributed to NSAID use. No unexpected DMARD toxicity or mortality occurred. Concomitant therapy for comorbidity, in particular for cardiovascular disease, osteoporosis, and gastrointestinal disease, increased: more than 60% were on these therapies at 20 year followup. CONCLUSION: Strategies to improve the outcome of RA in all dimensions should include: earlier referral for expert assessment; avoidance of NSAID gastrointestinal and nephrotoxicity; a more intensive effort to identify effective management of comorbidity and those likely to have a poor outcome. Such patients require sustained, intensive therapy to minimize later disability.