We have previously demonstrated that hydroxyurea (HU) reduces the rate of vascular complications in patients with essential thrombocythaemia (ET) at high risk of thrombosis. However, the relatively short follow-up (median 27 months) did not enable the evaluation of the risk of developing secondary malignancies. In this study, we report the long-term outcome of the 114 patients included in the trial: 56 patients randomized to receive HU and 58 patients to receive no cytoreductive therapy. Before randomization, 15 patients had been treated with busulphan. During the observation period, 29 patients (50%) shifted from the control to the HU group mainly because of thrombosis. Median follow-up was 73 months (range 3-94). Analysis was by intention to treat and, when indicated, by treatment. When analysed by intention to treat, 46 out of 54 patients (85%) originally randomized in the HU group are alive, compared with 49 of 58 patients (84%) in the control group [not significant (n.s.)]. Five patients (9%) in the HU group and 26 patients (45%) in the control group had thrombosis (P < 0.0001). Seven patients (13%) in the HU group developed secondary acute leukaemia, myelodysplastic syndromes or solid tumours, compared with only one of the control group patients (1.7%) (P = 0.032). The occurrence of secondary malignancies was also analysed by treatment: none of the 20 patients who had never been treated with chemotherapy developed neoplasia vs. three of the 77 patients given HU only (3.9% n.s.) and five of the 15 patients given busulphan plus HU (33% P < 0. 0001). This study showed that: (a) HU reduced the risk of thrombosis in ET patients; (b) the sequential use of busulphan and HU significantly increased the risk of second malignancies; and (c) overall survival was not affected by HU therapy.