Given the steep dose-response relationship with cisplatin, a pilot study on very high-dose cisplatin (HD-CDDP) was conducted in previously untreated patients with advanced ovarian carcinoma and postoperative residual tumor (RT). Thirty-seven patients (FIGO stages III-IV; RT> 0.5 cm) received three courses of HD-CDDP (a course of 40 mg m-2 day-1 for days 1-5, every 28 days). Twenty patients (54%) achieved clinical complete response (CR), 12 (32%) partial response (PR), and the remaining five (14%) showed stable or progressive disease (NC-P). All 20 clinically complete responders underwent second-look laparotomy and CR was confirmed in all but five cases (pathologic CR: 40%) and in 71% of patients with> 0.5-2 cm RT vs. 15% of those with> 2 cm RT (P < 0.001). The 4-year overall survival was 35% (median: 27 months, range: 7-58+), and 53% vs. 20% for patients with> 0.5-2 cm and> 2 cm RT, respectively (P = 0.01). The overall progression-free survival was 29.5% (median: 16 months, range 2-58+) and for patients with more or less than 2 cm RT it was 20 and 41.2% (P < 0.05). Pathologically complete responders received no further treatment and showed a 3-year disease-free survival of 53%. The major toxic effect was a delayed-onset peripheral neuropathy observed in all patients, five of them (13.5%) with gait disturbances requiring continuous assistance. Nevertheless, none of them became wheelchair dependent and about 90% of the alive patients recovered at the 18-month neurologic follow-up, suggesting that cisplatin damage can be reversible. Ototoxicity was detected in all patients although only 19% of patients were symptomatic. HD-CDDP showed high activity in patients with> 0.5-2 cm RT, suggesting that the adverse significance of minimal RT may be partially overcome through an intensive chemical cytoreduction. Substantial neurotoxicity and the need for intensive care represent the major drawbacks. Further studies should delineate the exact role of HD-CDDP in optimally debulked patients, and a considerable effort should be made in rapidly achieving reliable data on the value of neuroprotectors in the prevention of the dose-limiting neurotoxicity.