With the possible exception of chemotherapy for the small-cell type, a complete surgical excision is still the only effective treatment of lung cancer. Routine brain computed tomography (CT) for staging purposes has been both advocated and opposed. In this retrospective study, we aimed to assess the clinical yield of the technique. We saw 184 consecutive patients with a new histologically proven non-small-cell lung cancer. Using as reference criteria clinical judgment supported by a strict follow-up evaluation, we counted 1 false- and 23 true-positive brain CT results, plus 2 false- and 158 true-negative findings. These figures allow for sensitivity, specificity, and accuracy of 92 percent, 99 percent, and 98 percent. The frequency of brain metastases did not correlate with the various histologic types, even though adenocarcinoma was the most common cause of cerebral metastases. The absence of neurologic symptoms did not exclude cerebral involvement: in our experience, 16 of 25 patients with positive brain CT scans were asymptomatic (64 percent). Three of 31 subjects (10 percent) with an otherwise operable carcinoma were found to have metastases after brain CT. We conclude that routine cranial CT is useful in the staging evaluation of the patient with non-small-cell lung cancer (NSCLC) and that it should be performed in any candidate prior to surgical resection.