Controlled processing, response inhibition, and set adoption were examined in 51 HIV-1 infected participants and 21 uninfected controls who were administered a vocal reaction time (RT) version of the Stroop task (Stroop-RT; J. R. Stroop, 1935) as well as the traditional 100 item paper-and-pencil version. Response set expectancies on the Stroop-RT were manipulated by presenting 50% of trials in homogenous blocks and randomly varying the stimulus type during the remaining trials. As hypothesized, HIV seropositive (HIV+) participants were significantly slower than HIV seronegative controls on both versions of the Stroop. Significant interference effects were apparent on the paper-and-pencil version of the Stroop, but were not as prominent on the Stroop-RT. The HIV+ participants did profit from the blocking manipulation on the Stroop-RT, suggesting that set adoption is retained in HIV infection. These data suggest that HIV infection may result in deficient response inhibition, possibly secondary to frontostriatal dysfunction and dopaminergic alterations.