Two studies sought to determine whether perceived control has different effects on confidence assessment and betting decisions among pathological and problem gamblers than among non-problem gamblers. In Study 1, 200 college students who were frequent gamblers (80 female and 120 male, median age 20) completed the South Oaks Gambling Screen (SOGS) and then engaged in a task in which they answered questions, assessed confidence in each answer, and considered bets on their answers that were fair if they were well-calibrated, but unfavorable if they were overconfident. Probable pathological and problem gamblers earned significantly fewer points than non-problem gamblers. This was due to greater overconfidence among pathological and problem gamblers, which led to systematically less favorable bets. In Study 2, using 384 participants (105 female and 279 male, median age 20), control was independently manipulated and bets were constructed to make point value independent of overconfidence. Problem and pathological gamblers showed both greater overconfidence and greater bet acceptance. They were less affected by control in their betting decisions than non-problem gamblers, but more affected in the slope of their betting function. It is concluded that pathological and problem gamblers process information about confidence and control differently from non-problem gamblers.