Thirty-nine women with nonmetastatic gestational trophoblastic disease as determined by conventional staging studies were prospectively evaluated with computed axial tomography (CAT) of the lungs. Sixteen patients (41%) had pulmonary micrometastases detected by CAT, which were not detected by routine chest x-ray. Eight patients (20.5%) had indeterminate scans, and only 15 patients (38%) had negative scans. Eight of 16 (50%) with pulmonary micrometastases failed initial therapy with methotrexatefolinic acid rescue while one of eight (12.5%) patients in the indeterminate group and one of 15 (6.7%) patients in the true nonmetastatic group failed initial therapy (P < .006). All patients who failed methotrexate-folinic acid rescue ultimately achieved prolonged remission with actinomycin D. Time to remission was significantly decreased in patients without evidence of pulmonary micrometastases (P = .03), but the total number of courses of chemotherapy was not significantly different (P = .06). No life-threatening toxicity occurred. Pulmonary micrometastases detected by CAT but not chest x-ray are predictive of an increased risk of methotrexate-folinic acid therapy failure. Computed axial tomography of the lungs identifies a group of patients at high risk for failure of methotrexate-folinic acid rescue, and, therefore, may be indicated for routine staging of patients with otherwise nonmetastatic gestational trophoblastic disease.