The relationship of the Trail Making Test (TMT) to the frontal lobes was tested by comparing patients with damage to the frontal and nonfrontal regions to control participants. Although the analysis of time measurements, both raw and transformed, showed notable slowing of frontal groups, error analysis proved to be a more useful method of categorizing performance. Analysis of errors on Part B indicated that all patients who made more than 1 error had frontal lesions. Dividing the frontal damaged patients into subgroups on the basis of the number of errors yielded specificity of brain-behavior relations within the frontal lobes. Patients with damage in dorsolateral frontal areas were most impaired. Those with inferior medial damage to the frontal lobes were not significantly affected in TMT Part B performance.