Exploratory research into the role of visual and verbal work-ing memory in diagrammtic reasoning is presented. Eighty subjects participated in an experiment where thirty-four dif-ferent gain/loss problems were represented in sentential and graphical form. In order to test the role of each component of Baddeley’s model of working memory, subjects performed verbal, visual, and mental suppression tasks while reason-ing about sentential and graphical gain/loss problems, yield-ing six different conditions. In two additional control condi-tions, no suppression tasks were performed. Interference and preference reversals occurred in all six conditions involving suppression tasks, although no interference was predicted in conditions involving the graphical representation and verbal suppression task, or the sentential representation and visual suppression task. In the control conditions subjects made re-sponses consistent with prospect theory with little interfer-ence. The data suggests that certain gain/loss problems re-quire both visual and verbal resources and that the taxing of these resources results in strategies that may favor a minimal inference strategy.