176 eligible patients with advanced suboptimally operated ovarian carcinoma were randomly allocated to receive either cisplatin 75 mg/m2 or cisplatin 50 mg/m2 and cyclophosphamide 500 mg/m2 (CP) every 28 days for six courses. The overall clinical response rates (complete response plus partial response) were 52 and 63% for CP and cisplatin, respectively (non-significant). Including results obtained by second-look laparotomy, we did not observe a statistically significant difference in response rates in the two treatment groups. Median progression-free survival was 10 and 11.9 months for CP and cisplatin, respectively (non-significant). No significant difference was observed in overall survival, with a median of 19.4 and 21.5 months for CP and cisplatin, respectively. Thirty-seven platinum-resistant and 27 platinum-sensitive tumours were treated with carboplatin or cisplatin as second-line therapy. Response rates to platinum second-line therapy were 6 and 50% for resistant and sensitive tumours, respectively (P < 0.001). This difference in response rate was also confirmed by survival analysis. Patients with platinum-sensitive tumours survived longer when they were treated with platinum-containing chemotherapy (P = 0.005). Median survival was 22.8 and 8.5 months after initiation of second-line treatment for the platinum-containing and platinum-free regimens, respectively. In summary, we observed in suboptimally operated ovarian carcinoma patients similar response rates, progression-free interval, and overall survival for equitoxic cisplatin and CP. However, the doses of cisplatin and cyclophosphamide chosen were substantially lower than current standard doses of CP. Our study demonstrates, therefore, that a suboptimal dose of CP is as effective as optimal dose monotherapy cisplatin. Patients with recurrences considered as platinum-sensitive had a significantly higher response rate and improved survival when retreated with platinum-containing therapy.