We investigated the accuracy of both lymphography and computed tomography (CT) in detecting lymph node metastases in 58 patients with primary epithelial ovarian cancer (group a) and subsequently submitted to surgery including pelvic and/or lumbo-aortic lymphadenectomy. CT accuracy was also investigated in 41 patients with clinically suspected relapse of ovarian cancer (group b). In the first group (a) overall results in the pelvis were, respectively, for lymphography and CT: 94.8 vs 89.6% accuracy, 85.7 vs 57.1% sensitivity, 97.7 vs 100% specificity, 97.7 vs 100% positive predictive value and 95.5 vs 88% negative predictive value. In the lumbo-aortic region, we had: 88.8 vs 86.1% accuracy, 71.4 vs 64.2% sensitivity, 100% specificity and positive predictive value for both techniques, and 84.6 vs 81.6% negative predictive value. In the second group (b) CT accuracy, sensitivity and specificity were 90.2, 80 and 100%, respectively. CT, thanks to its high specificity and positive predictive value, can represent the method of choice for the evaluation of pelvic and lumbo-aortic lymph node metastases in untreated and relapsing ovarian cancer. CT demonstration of lymph node metastases can affect not only lesion staging, but also chemotherapy; different indications for lymphadenectomy may also depend on CT.