When receiving disconfirmatory social feedback about recollected events, people sometimes defend and sometimes reduce their belief that the event genuinely occurred. To improve estimates of the rates of memory defense and reduction, and of the magnitude of the change in belief in occurrence that results, in the present studies we examined the effect of disconfirmatory social challenges made to correctly recalled memories for actions performed in the lab. Adult participants performed, imagined, or heard action statements and imagined some of the initial actions multiple times. One week later, they completed a source-monitoring test and rated the actions on belief in their occurrence, recollection, visual detail, vividness, and reexperiencing. Four of the correctly recalled performed actions were challenged either prior to making the ratings during the test (Study 1, N = 44) or after making initial ratings after completing the test, following which the ratings were taken again (Study 2, N = 85). Across both studies, challenges were associated with lower belief-in-occurrence and recollection ratings on average than for control items, and belief in occurrence was affected to a greater extent than recollective features. Challenges that occurred during the test produced more instances of defense, whereas challenges that occurred after the test produced more instances of reduction. A closer analysis showed that some participants always defended, some always reduced, and some both defended and reduced belief. Responses to the first challenge positively predicted the responses to subsequent challenges. In addition, the procedure in Study 2 produced a variety of types of nonbelieved memories.