Systematic restaging was performed prospectively in 100 patients referred to the Ovarian Cancer Study Group institutions with a diagnosis of "early" (stage Ia-IIb) ovarian cancer. Before referral, only 25% of patients had an initial surgical incision that was adequate to allow complete examination of the pelvis and abdominal cavity. In patients referred to member institutions, 31 (31%) of 100 were found to have a more advanced stage and 23 (77%) of 31 of these actually had stage III disease. Sixty-one percent of the patients had their advanced stage detected by procedures other than a second laparotomy-nine (29%) of 31 by peritoneoscopy, six (19%) of 31 by peritoneal washings, and six (19%) of 31 by lymphangiography. Sites of unsuspected disease are most likely to be pelvic peritoneum, ascites fluid, other pelvic tissue, para-aortic nodes, and the diaphragms. Based on these data, we conclude that the initial staging approaches traditionally used in clinical evaluation of patients with early ovarian cancer are often incomplete and inadequate.