The long-term use of second-line antirheumatic drugs was prospectively studied in a consecutive sample of 245 patients with recently diagnosed rheumatoid arthritis. A survival analysis was done in which treatment termination due to side-effects and to insufficient therapeutic effect were used as index causes. Cumulative drug 'survival' of aurothioglucose with treatment termination due to toxicity was significantly less compared with hydroxychloroquine. With regard to lack of efficacy as index cause, the administration time of hydroxychloroquine was significantly less than that of either aurothioglucose or sulphasalazine. Treatment termination due to lack of efficacy or combined insufficient therapeutic response and toxicity proved to be influenced by the initial disease activity and by the rank order of prescription.