An extended phase II study was performed to evaluate single-agent paclitaxel as salvage chemotherapy for ovarian cancer. The aim of this study was to evaluate the 3-h infusion schedule of paclitaxel in terms of toxicity and antitumour efficacy. Furthermore, we analysed the impact on response and survival of the extent of prior chemotherapy and status of resistance against platinum. This study was an open, non-randomised, multicentre trial. The dose of paclitaxel used was 175 mg/m2 in patients who had received one or two prior therapies, and 135 mg/m2 in patients who had received three prior therapies. Paclitaxel was given as a 3-h infusion. Courses were repeated every 3 weeks. 114 patients with platinum-pretreated epithelial ovarian cancer were recruited of whom 112 were found eligible and evaluable for toxicity. 104 patients with bidimensionally measurable disease who received more than one course of chemotherapy were evaluable for response, progression-free (PFS) and survival. Toxicity was generally manageable. Main toxicities were non-cumulative neutropenia with 22.3% of courses with WHO grade 3/4 and peripheral neuropathy which occurred in more than half of the courses and was of WHO grade 2 and 3 in 20.1 and 1.3% of the courses, respectively. Neuropathy was associated with the higher dose per course and with cumulative paclitaxel dose. Objective responses were reported in 20% (21/104) of the patients (95% CI 13-29%) with a median response duration of 36.7 weeks. Survival and PFS for the whole group were 45.9 and 15.1 weeks, respectively. Performance status, number of tumour lesions and extent of prior chemotherapy were found to be prognostic factors for survival. Extent of prior chemotherapy was the only prognostic factor for PFS. Platinum resistance did not predict response to treatment. Paclitaxel 175 mg/m2 given as a 3-h infusion is an appropriate treatment for patients with platinum-resistant ovarian cancer who have not previously received more than two chemotherapy regimens. Paclitaxel did not show results superior to historical data for platinum retreatment in patients with platinum-sensitive, recurrent ovarian cancer.