The findings from preoperative diagnostic ultrasound were compared with operative and pathologic findings in patients with ovarian cancer. Twenty laparotomies were diagnostic second-look procedures and 21 were therapeutic procedures in patients with clinically evident disease. Ultrasound was more reliable in patients with clinically suspected disease, but a site by site analysis demonstrated it is insensitive for detecting prevertebral adenopathy less than 3 cm in size, thin, 1.5 cm omental plaques, 5 cm or smaller lesions involving the mesentery or bowel and peritoneal masses 2 cm or less in size. Ultrasound was most sensitive in the detection of ascites.