The Iowa Gambling Task (IGT) simulates uncertain gains and losses in real life situations and thus is a good measure of uncertain decision making. The role of working memory (WM) in IGT performance still remains unclear. The present study aimed to examine the effect of WM on IGT performance. Three groups of participants matched on gender ratio were randomly assigned to no WM load, low WM load, and high WM load conditions. Initially the three groups did not show significant difference (...) in WM capacity. They finished a modified version of IGT and then their implicit learning effect and explicit cognition on IGT were assessed. Results indicated a linear increasing trend of IGT performance among high WM load, low WM load and no WM load groups; participants in the no WM load and low WM load groups revealed implicit learning effect, while participants in the high WM load group did not; all participants showed explicit cognition on IGT to the same level. These results suggested that participants in the high WM load group showed good explicit cognition to IGT but showed poor performance. This pattern is similar to frontal patients. Further studies should be conducted to explore this issue. (shrink)