In recent years, interest in the use of computerized neuropsychological (NP) assessment measures has increased. However, there are limited data regarding how performance on these measures relates to performance on more traditional, clinical instruments. In the present study, 82 HIV+ men, who were all believed on clinical grounds to have neurobehavioral impairment, completed a traditional NP battery (TNB) and the California Computerized Assessment Package (CalCAP, a collection of computerized reaction time tests). Summary scores based on a TNB, as well as those based on the CalCAP, demonstrated significant associations with both degree of immunosuppression (CD4 count) and detectable viral load in cerebrospinal fluid, but not with detectable viral load in plasma. Established norms on the TNB and CalCAP batteries resulted in classifying 57% and 49% of the HIV+ sample as impaired, respectively. When using the TNB as the "gold standard," impairment classifications based on CalCAP summary scores exhibited a sensitivity of 68% and a specificity of 77%. Overall agreement on impairment classifications between batteries was low (kappa = .44). Data from this study suggest that traditional NP batteries and computerized reaction time tests do not measure the same thing, and are not interchangeable in examining HIV-related NP impairments.