Individuals tend to be overconfident when making retrospective judgments about the quality of their decisions. However, few studies have focused on age differences in estimates of decision quality. In the present experiment performance estimates were provided by task-trained and untrained young and old individuals following completion of a series of complex financial decisions. Confidence levels were assessed by examining discrepancies between perceived and actual solution quality. Performance estimates of all 4 groups contained appreciable estimation error; however, no group showed a substantial directional bias toward underconfidence or overconfidence. Young trainees were significantly less confident in the quality of their decisions than young novices, but a comparable training effect was not found among older individuals. One's knowledge of the task, prior decision-making experience, and level of self-esteem may combine to determine the accuracy of one's retrospective performance estimates.