Although improvements in performance due to TMS have been demonstrated with some cognitive tasks, performance improvement has not previously been demonstrated with working memory tasks. In the present study, a delayed match-to-sample task was used in which repetitive TMS (rTMS) at 1, 5, or 20 Hz was applied to either left dorsolateral prefrontal or midline parietal cortex during the retention (delay) phase of the task. Only 5 Hz stimulation to the parietal site resulted in a significant decrease in reaction time (RT) without a corresponding decrease in accuracy. This finding was replicated in a second experiment, in which 5 Hz rTMS at the parietal site was applied during the retention phase or during presentation of the recognition probe. Significant speeding of RT occurred in the retention phase but not the probe phase. This finding suggests that TMS may improve working memory performance, in a manner that is specific to the timing of stimulation relative to performance of the task, and to stimulation frequency.