Recent studies show that visual search is less efficient when loads are placed on working memory (WM) by having subjects manipulate information (i.e. count backwards from a target digit) but not when subjects simply maintain information in WM (i.e. rehearse a set of random digits). We tested the effects of different WM loads (maintain versus manipulate) on driver detection of peripheral letter targets and on driving performance using a driving simulator. Participants (n=40) performed secondary tasks that loaded either visuo-spatial working memory or verbal working memory (phonological loop). The verbal task required participants to either rehearse a string of random letters or to alphabetize the string while driving. In the visual task the participants were presented with an image of a human stick figure in different orientations (up-right, up-side down, facing forward or facing backward) and holding different geometric shapes in each hand. The participants were asked to either remember the figure or to identify what the figure was holding in one of the hands. The simulation consisted four equal length segments demarcated by stop lights. At the start of each road segment the participants were presented with a random letter string or a stick figure and asked to memorize the stimulus or to manipulate it (i.e, alphabetize or identify which hand held a particular shape) while driving. The results show that subjects detected fewer peripheral letter targets (p =.016) and responded more slowly to peripheral targets (p=.002) when manipulating information using WM regardless of modality (visual or verbal). Subjects also received more speeding tickets but drove slightly slower while engaged in the manipulation tasks. These data show that WM tasks that require participants to manipulate information regardless of modality (visual or verbal) negatively affect driving performance and interfere with a drivers' visual search of the driving environment.