The surgical staging scheme for uterine corpus cancer adopted in 1988 by the International Federation of Gynecology and Obstetrics assigns patients with tumor spread to retroperitoneal lymph nodes to stage IIIC. However, a recommended approach to the detection of lymph node metastasis is not delineated. As part of an ongoing project to assess the value of surgical staging procedures, we reviewed the techniques of lymph node evaluation in 295 at-risk patients. Cases included clinical stage I patients whose preoperative biopsies demonstrated grade 2 or 3 adenocarcinoma or papillary serous, clear cell, or mixed carcinoma. We arbitrarily divided the retroperitoneal space into 10 lymphatic zones: left and right para-aortic, common iliac, external iliac, hypogastric, and obturator. Eighty-two percent of patients had some type of node sampling that involved a mean of three zones. Thirty-three of 244 sampled cases (13.5%) had nodal metastases: 20 had gross involvement and 13 had microscopic. We stratified patients into three groups: (1) those who had no node sampling (n = 51), (2) those with some nodes biopsied (n = 193), and (3) those whose node sampling included a minimum of one para-aortic plus at least one right and left pelvic specimen (n = 51). Retroperitoneal recurrences thought to originate from lymph node sites were identified for the "node-negative" patients in each group: Group 1, 4/51 (8%); Group 2, 9/173 (5%); and Group 3, 0/38 (0%). Lymphatic site failures were seen in 8 of 33 (24%) patients with biopsy-proven metastases. We found that failure to systematically sample pelvic and para-aortic nodes results in a small, but real, risk of undetected extrauterine metastasis. A selective approach to sampling that includes biopsy from both para-aortic and bilateral pelvic lymphatic zones appears to provide an accurate estimate of true node negativity. Further evaluation of this approach is warranted.