Subtypes of working memory performance were examined in a cohort of 50 HIV-infected adults and 23 uninfected controls using an n-back paradigm (2-back) in which alphabetic stimuli were quasi-randomly presented to a quadrant of a computer monitor. In the verbal working memory condition, participants determined whether each successive letter matched the letter that appeared two previously in the series, regardless of spatial location. In the spatial working memory condition, participants determined whether each letter matched the spatial location of the letter that had appeared two previously, regardless of letter identity. The dependent variable was percent accuracy in each condition. Results of mixed model ANOVA revealed that the HIV-infected participants performed significantly worse than controls on both the verbal and spatial working memory tasks. A significant main effect for working memory condition was also present with both participant groups performing better on the spatial working memory task. These results, the first study of HIV-infected adults to directly compare verbal versus spatial working memory performance using the identical test stimuli across task conditions, suggests that HIV infection is associated with a decrement in working memory efficiency that is equally apparent for both verbal and spatial processing. These findings implicate central executive dysfunction as a likely substrate and provide the basis for hypothesizing that decline in working memory may contribute to other HIV-associated neuropsychological deficits. (JINS, 2002, 8, 532–538.)