Women with stage III ovarian cancer and with < or = 2 cm residual tumour were randomly assigned to receive either conventionally dosed chemotherapy (group A) or HDCT (group B). Patients allocated to group A received 6 cycles of paclitaxel (T) 135 mg/m2 and cisplatin (P) 75 mg/m2 every 3 weeks, and those allocated to HDCT received 3 TP cycles followed by peripheral blood stem cell mobilisation with cyclophosphamide (C) 3000 mg/m2 and T 175 mg/m2, and subsequently HDCT with carboplatin 1500 mg/m2, C 120 mg/kg, and mitoxantrone 75 mg/m2. The trial was closed early after 42 patients were entered due to slow accrual. The median follow-up time of patients who were alive was 81 months. The median progression-free survival time was 15.9 and 16.6 months (hazard ratio, HR 0.83; 95% CI 0.41-1.69, P = 0.61) and the median overall survival time was 43.7 and 64.3 months (HR, 0.74; 95% CI 0.34-1.61, P = 0.44) in groups A and B, respectively. Although one patient died of HDCT-related toxicity, the regimen was otherwise relatively well tolerated. We conclude that the HDCT regimen used was feasible, but did not result in significantly improved survival in this prematurely closed trial. A clinically important survival benefit cannot be excluded due to the small sample size.