Abstract The efficacy of frozen section in detecting metastases in pelvic and or periaortic lymph nodes during radical hysterectomy is unknown. The finding of positive nodes may result in termination of the operative procedure. In this study, we attempted to determine the accuracy of frozen sections in this situation. Intraoperative pathology consultation records were examined for 127 patients undergoing surgical exploration for radical hysterectomy between 1977 and 1992. Microscopic slides of lymph nodes were reviewed for accuracy. Metastasis diameters were measured and blocks cut five close microtome levels deeper. In 19 cases (15%) positive nodes were documented on permanent section, with metastases ranging in size from less than 1 to 19 mm. Thirteen cases of node metastasis were diagnosed at frozen section. All were suspicious to the pathologist on palpation and gross inspection after bisection. Six cases were missed by sampling error on frozen section; in 4, metastases were smaller than 1 mm; in 1, between 2 and 3 mm; and in 1, 19 mm. The sensitivity was 68%, the false-negative rate was 32%, and the specificity was 100%. No cases were false positive at frozen section. The frequency of nodal metastasis and detection rate by frozen section did not differ significantly between carcinoma types. No micrometastases (<2 mm) were detected by frozen section. All micrometastases were no longer present within five microtome levels. Frozen section diagnosis of pelvic node metastasis is a highly specific procedure which should alter intraoperative management of early-stage cervical cancer.