Because crack cocaine appears to have a preferential effect on the metabolic and electrophysiological activity of the frontal and temporal regions of the brain (Pascual-Leone et al., 1991a, 1991b; Volkow, 1992), we hypothesized that cognitive measures of those regions would be impaired in crack cocaine users relative to measures in normal volunteers. We used logistic regression to determine the relationship of cocaine usage to neuropsychological test performance. We compared 38 patients with an average of 3.6 (SD = 2.5) years of crack cocaine use and 24.5 (SD = 28.1) days of abstinence to 54 normal volunteers on a battery of neuropsychological tests. Statistical adjustments were made for the effects of age, education, socioeconomic class, and level of depression. Our findings were mixed with regard to purported measures of executive/frontal functioning, with worse performance associated with cocaine usage on the Booklet Categories Test, but better performance associated on others (number of categories on the Wisconsin Card Sorting Test, Controlled Oral Word Association). Cocaine usage was associated with impairment on measures of spatial, but not verbal memory, confrontation naming, and Trail-making Test, Part B, a measure of perceptual-motor speed and cognitive flexibility. In summary, it appears that continuous crack cocaine use produces a dissociative pattern in neuropsychological test performance with improvement on some measures, but deterioration on others. The permanence of these effects remains to be determined with longitudinal studies.