We performed a prospective, parallel, descriptive study of 737 consecutive new uses for 11 drugs prescribed for patients with definite or classic rheumatoid arthritis. The patients were from 5 geographically dispersed sites. Researchers used validated outcome assessment instruments to measure endpoints of disability, pain, patient global assessment, medication costs, laboratory costs, and number of physician visits. Patients were studied by strict prospective protocol at 6-month intervals for 3 years. Controls included parallel results with other drugs, and before and after values for the individual patient. Beneficial effects were observed with the "disease-modifying" drugs: intramuscular gold, penicillamine, and methotrexate. Of these, gold had the most apparent effect. An average of 9 months of gold therapy resulted in highly significant reductions in disability (P less than 0.005), pain (P less than 0.001), and patient global assessment (P less than 0.005). However, patients receiving gold and methotrexate had nearly twice as many visits to physicians. In addition, drug costs increased strikingly with gold, and laboratory costs tripled. Relatively minor differences among nonsteroidal antiinflammatory agents were difficult to interpret. The outcome assessment techniques used in the study are sensitive measures, which confirm the results of experimental studies and extend observations to new outcomes, including cost and disability.