The aims of this study were to determine which cognitive domains are evaluated by the Paced Auditory Serial Addition Test (PASAT) and to identify which of them are responsible for the poorer performance displayed by multiple sclerosis (MS) patients in this task. A total of 30 healthy controls and 30 MS patients completed the PASAT task as well as the different tests contained in the Brief Repeatable Battery of Neuropsychological Tests (BRB-N), some Wechsler Adult Intelligence Scale (WAIS-III) subtests, the Spanish version of the Chicago Multiscale Depression Inventory (CMDI), and a new PASAT-based task (ADD1) that was specifically designed for this study. Analysis of covariance and regression-based analyses were performed to identify the predictors that are most strongly associated with the PASAT scores and the between-groups differences in the performance of this task. PASAT execution was associated with scores of the Digit Backward test, Symbol Digit Modality Test (SDMT), and measures of working memory and information-processing speed. On the other hand, differences between healthy volunteers and MS patients were mainly associated with the SDMT scores. MS patients also exhibited poorer execution than controls in the ADD1 task. Our results suggest that reduced information-processing speed (and not working memory) is the primary alteration underlying the lower scores in the PASAT task (and probably other cognitive deficits) that characterize MS patients. Based on these results, we suggest that tests that capitalize the role of information-processing speed may be of special relevance in the neuropsychological assessment of this clinical population.