OBJECTIVE To study the long term tolerance of parenteral gold and subsequent drug treatment in patients with rheumatoid arthritis, including prediction of outcome and 'survival' of sequential treatments. METHODS A retrospective cohort study of 376 patients was made, including a detailed screening of 237 patients treated in 1989. Reasons for discontinuing treatment were analysed in life table analyses, which were used to compare patients receiving parenteral gold treatment in 1985 and 1989, and two groups of patients receiving disease modifying antirheumatic drugs after parenteral gold treatment. The causes of discontinuation were followed in sequential treatments. RESULTS The estimated probability of discontinuation of parenteral gold treatment was 29% after six months and 42%, 55%, 74%, and 92% after 1, 2, 5, and 10 years, respectively. Mucocutaneous side effects were the main cause of discontinuation of parenteral gold treatment during the first three years, while the probability of discontinuation because of inefficacy dominated after four years. Side effects also constituted the main cause of discontinuation of treatments given after parenteral gold treatment during the first three years of follow up. No significant differences were found when comparing the termination rates between the first and the second and subsequent treatments after parenteral gold treatment. The main reasons for discontinuing one treatment could not predict the cause of discontinuation of the next treatment. CONCLUSIONS Mucocutaneous side effects dominated initially, while inefficacy was the dominating cause of discontinuation of long term parenteral gold treatment. No serious side effects were registered. The cause of discontinuation of one treatment did not predict the cause of discontinuation of the following drug. Drug 'survival' was the same in both treatments after parenteral gold treatment.