A long-term retrospective case-control study was performed comparing 119 patients with rheumatoid arthritis treated with cyclophosphamide and 119 matched control patients with rheumatoid arthritis not treated with cyclophosphamide to determine the risk of subsequent malignancy. Thirty-seven malignancies were detected in 29 cyclophosphamide-treated patients, while 16 malignancies were found in 16 control patients (p less than 0.05) during a mean follow-up period of more than 11 years. Urinary bladder cancer (six cyclophosphamide-treated patients, no control patients) and skin cancer (eight cyclophosphamide-treated patients, no control patients) were identified as differing statistically between the groups, and hematologic malignancy (five cyclophosphamide-treated patients, one control patient) showed a similar trend. Survival analysis indicated that the rate of development of malignancy in the cyclophosphamide-treated patients was significantly greater than in the control patients at six years following drug initiation, and that this increased rate persisted even at 13 years (p less than 0.01). Of the many risk factors evaluated, mean total cyclophosphamide dose and duration and tobacco use were significantly increased in patients in whom cancer subsequently developed. These long-term complications must be considered seriously when cyclophosphamide or other cytotoxic drugs are initiated for the treatment of rheumatoid arthritis.