We report an observational study of chemotherapeutic regression of ovarian tumor implants according to decrements in residual mass size after surgical cytoreduction. Cytoreductive operations were attempted on 74 consecutive patients with stages IIIB-IV disease referred for this purpose. Thirty-two patients had received one to four courses of preoperative chemotherapy (22 responses, no progressions). Postoperative chemotherapy followed current protocols at Dana Farber Cancer Institute (n=61) or referring institutions (n=13); 57 regimens contained cisplatin. Postchemotherapy response was assessed clinically or by second-look procedures. Negative findings were considered a complete remission. Masses >1 cm were excised from 62 patients. Twelve patients were inoperable. Twenty-eight patients had complete remissions and the correlation between these and decrements in residual mass size was highly significant (P <0.0001). Complete remissions had a uniform effect and were the only outcome predictive of survival. Preoperative treatment greatly facilitated cytoreduction but only masses 0-0.2 cm were senstitive to postoperative chemotherapy. Masses 0.5 cm or less were optimal. They made up 77% of operable patients and supplied 25 (89%) of the complete remissions. Cytoreduction is not always required but even large-volume disease in the upper abdomen can be safely excised. The concept that masses larger than 10 cm indicate general chemoresistance has not been sustained.