Lipiodol injected into the hepatic artery is selectively retained in hepatomas so has been used as a vehicle for cytotoxic drugs. This study compared treatment with 5-epidoxorubicin emulsified in lipiodol and infused into the hepatic artery with symptomatic treatment alone in a randomised trial. Of 136 patients with hepatoma 78 (57%) were not eligible, eight (6%) refused to take part, and 50 entered the trial (chemotherapy: n = 25, symptomatic treatment: n = 25). The two groups had similar prognostic indices. Seven of 25 patients allocated to chemotherapy were unable to receive it. The slight survival disadvantage associated with chemotherapy was not significant (median survival 48 days compared with 51 days, log rank chi 2 = 0.07, p > 0.05). Patients given chemotherapy spent significantly longer in hospital, however (median three days compared with one, p = 0.0008). Changes in symptoms and indices of tumour growth did not differ significantly between the two groups. It is concluded that infusion of 5-epidoxorubicin emulsified in lipiodol for hepatoma increased morbidity but did not affect survival. In addition, most patients were unsuitable for this treatment because of advanced disease. The patients in the trial had a short median survival time so the conclusions may not be valid for other patients with hepatoma.