A multicentre, randomised study was carried out in Europe, South Africa and North America to compare the activity and tolerability of oral versus intravenous (i.v.) topotecan in patients with relapsed epithelial ovarian cancer. Patients who had failed first-line therapy after one platinum-based regimen, which could have included a taxane, were randomised to treatment with either oral (p.o.) topotecan, 2.3 mg/m(2)/day or i.v. topotecan 1.5 mg/m(2)/day for 5 days every 21 days. Patients were stratified by prior paclitaxel exposure, interval from previous platinum therapy and tumour diameter. 266 patients were randomised. Response rates were 13% orally (p.o.) and 20% (i.v.) with a complete response in 2 and 4 patients, respectively. The difference in the response rates was not statistically significant. Median survival was 51 weeks (p.o.) and 58 weeks (i.v.) with a risk ratio of death (p.o. to i.v. treatment) of 1.361 (95% confidence interval (CI): 1.001, 1.850). Median time to progression was 13 weeks (p.o.) and 17 weeks (i.v.). The principal toxicity was myelosuppression although grade 3/4 neutropenia occurred less frequently in those receiving oral topotecan. Toxicity was non-cumulative and infectious complications were relatively infrequent. Non-haematological toxicity was generally mild or moderate. The incidence of grade 3/4 gastrointestinal events was slightly higher for oral than i.v. topotecan. Oral topotecan shows activity in second-line ovarian cancer and neutropenia may be less frequent than with the i.v. formulation. A small, but statistically significant, difference in survival favoured the i.v. formulation, but the clinical significance of this needs to be interpreted in the context of second-line palliative treatment. Oral topotecan is convenient and well tolerated and further studies to clarify its role are ongoing.