Computerized reaction time (RT) tasks are sensitive measures of subclinical HIV-related mental slowing. We previously reported that nondemented HIV-seropositive patients on antiretroviral therapy at the time of testing had faster choice RTs compared to matched untreated seropositive participants. In the present study, we evaluated the performance of 163 nondemented HIV-seropositive participants on a reaction time version of the Stroop task as a function of antiretroviral status. Persons on antiretroviral therapy at the time of testing had significantly faster reaction times than untreated individuals, although treated asymptomatic participants showed significantly less Stroop interference than treated symptomatic participants. These effects could not be attributed to differences in demographic variables, disease status, substance abuse, or psychological distress. These data indicate that central information processing is faster for patients treated with antiretroviral compounds compared to untreated patients, and suggest that reaction time tasks may have significant potential utility in clinical trials of neuroprotective compounds.