A collaborative group of cancer registries and hospitals carried out a case—control study of tumours of the bladder in women who had previously been treated for ovarian cancer. A total of 63 cases of bladder tumours were identified, and 188 controls were selected matching for age, year of ovarian cancer diagnosis and survival time. Full details of the treatment for ovarian cancer were sought both for cases and for controls. The risk of bladder tumours was increased for patients who had been treated by radiotherapy alone (1.9; 95% confidence interval, 0.77–4.9), by chemotherapy alone (3.2; 0.97–10), and by chemotherapy and radiotherapy (5.2; 1.6–16), when comparison was made with patients treated only by surgery. Patients treated by chemotherapy were separated into 2 groups according to whether they had received cyclophosphamide. Among those who had, there was a clear increase in risk (approximately 4-fold) regardless of whether or not they had also received radiotherapy. For those who received only other drugs, risk was increased substantially among patients who had also been treated by radiation, as compared with patients treated by surgery alone, and those who had received radiotherapy only. Both melphalan and thiotepa were implicated as potential bladder carcinogens on the basis of these results. The estimated risk of bladder tumours due to cyclophosphamide was more than twice the risk following radiation to the bladder, and it appeared substantially earlier. For both agents, the risk continued to increase more than 10 years after treatment began.