Three hundred twenty patients were entered into GOG Protocol 63, a clinical-pathologic study of stage IIB, III, and IVA cervical carcinoma. Following the completion of FIGO staging prerequisites, patients had computerized tomography (CT), a lymph-angiogram (LAG), and an ultrasound (US) of the aortic area. If any study was positive, a cytologic or histologic evaluation by fine-needle aspiration or selective paraaortic lymphadenectomy was performed. Paraaortic node dissection was mandated for patients with negative extended staging studies. Results of extended staging evaluations were compared with histologic or cytologic results. Two hundred sixty-four patients were eligible and evaluable. One hundred sixty-seven patients (63%) were stage IIB, 89 (34%) were stage III, and 8 (3%) were stage IVA. Positive paraaortic nodes occurred in 21% of stage IIB, 31% of stage III, and 13% of stage IVA. LAG sensitivity was 79% with a specificity of 73%. Sensitivity of CT and US was 34 and 19%, respectively, with specificities of 96 and 99%, respectively. The frequency of false-negative results with LAG for patients with stage IIB disease was 6%. This decrease is consistent with a stable sensitivity and specificity. These findings suggest that a negative LAG may be adequate to eliminate surgical staging in subgroups with low risk of metastasis to the aortic nodes. Until new noninvasive testing methods are developed, LAG appears to be the most reliable noninvasive examination to evaluate spread of cervical cancer to aortic nodes.