Reappraisal of disease-modifying antirheumatoid drug (DMARD) therapy in rheumatoid arthritis (RA) has raised the possibility that the risks of treatment outweigh the benefits. To provide more information a retrospective survey of the case-notes of 325 patients with RA was performed. The case-notes were randomly selected from the 2320 RA patients indexed in one department. Improvement was defined by the contemporary written notes of the managing physician. Analysis employed the life-table technique. A total of 247 case-notes could be adequately analysed: 154 patients had received one or more DMARDs constituting 251 drug-patient exposures. Improvement followed drug-patient exposure in 57% of cases after a delay of between one and seven months. Nine per cent occurred within one month and 22% within two months. Of those patients withdrawn from treatment after less than two months, the 'early withdrawal' group, 25% subsequently improved without further DMARD therapy. The probability of still receiving a specific DMARD 8 months, 24 months and 36 months after the start of treatment was 50%, 25% and 10%, respectively. The majority of withdrawals resulted from adverse reactions. There was no evidence for a relationship between patient responses to sequential DMARDs.