HIV-seropositive and seronegative control subjects performed a standard paper-and-pencil version and an experimental reaction time version of the Stroop Color-Word Naming Task. Results indicated that both symptomatic and asymptomatic HIV-positive subjects showed an exaggerated Stroop effect compared to controls, but this increase was only apparent on the RT version of the task. Analysis of components of the effect indicated that HIV-positive subjects showed increased inhibition compared to controls but normal facilitation. These results suggest that HIV-related cognitive slowing has an attentional component, most likely involving controlled processes. In addition, these results emphasize the utility and sensitivity of RT measures in the study of early HIV-1 infection.