The HIV dementia scale (HDS) previously has been shown to be sensitive in identifying persons with HIV-associated dementia (HAD). We aimed to develop a simpler version of the HDS--the Modified HDS (M-HDS)--which would be appropriate for use by nonneurologists. We sought to determine whether the M-HDS and the widely used Grooved Pegboard (GP) were comparable in identifying and staging the severity of dementia in HIV-seropositive persons. Data from 455 HIV-seropositive persons were analyzed, and patients were separated into 2 groups: demented (n = 144) and nondemented (n = 311), where the diagnosis was established by a single neurologist. Of the 144 with HAD, 13 had severe dementia, 55 had moderate dementia, 51 had mild dementia, and 25 had minor cognitive/motor disorder. We assessed the relationship between dementia severity and score on the M-HDS and GP (nondominant hand). Results showed both the M-HDS (z = -4.32; P < .05) and the GP nondominant (z = -4.08; P < .05) differentiated demented from nondemented persons. Furthermore, the M-HDS (z = -2.95; P < .05) and GP nondominant (z = -2.68; P < .05) discriminated equally well the stages of dementia severity. A cutoff score of 7.5 on the M-HDS maximized its sensitivity (70%) and specificity (71%). The M-HDS and the GP (nondominant) were found to be equally sensitive and specific in categorizing and staging HAD. However, the M-HDS requires no equipment and is simpler to administer; so it may be more useful for screening by nonneurologists.