We examined the sensitivity and specificity of the Wisconsin Card Sorting Test (WCST) as a measure of frontal lobe damage in 91 subjects with stable focal brain lesions. Anatomical information about the location and extent of brain damage was obtained from MR and CT transparencies. No significant differences in WCST performance were found between subjects with frontal vs. nonfrontal damage. Some subjects with extensive frontal lobe damage performed well on the WCST, and some subjects with damage outside of the frontal lobes failed. The optimal cutoff scores for discriminating frontal from nonfrontal subjects correctly classified only 62% of the subjects. Further analysis of WCST performances associated with damage to various subregions of the frontal lobes also failed to reveal any reliable relationships. These findings indicate that performance on the WCST cannot be interpreted in isolation as an index of frontal lobe damage.