Scales were constructed to measure perceived control over controllable events (realistic control) and perceived control over uncontrollable events (unrealistic control). Internal reliability, test-retest reliability, and discriminant validity of both scales were adequate. Study 1 measured perceived personal control over hassles that judges rated on general controllability. For hassles very high in controllability, perceived personal control was related to belief in realistic control but not to belief in unrealistic control; for hassles very low in controllability, perceived personal control was related to belief in unrealistic control but not to belief in realistic control. Study 2 showed that participants high in unrealistic control belief (but not those high in realistic control belief) persevered more on a task that was in part uncontrollable. Study 3 showed that the combination of low realistic control belief and high unrealistic control belief predicted poorer future health, particularly for participants who have reported the experience of many negative events and/or hassles. The conditions under which unrealistic control results in maladaptive outcomes are discussed.