Objective: To study whether memory control beliefs predict response to memory training, or change as a result of participating in memory training. Methods: Eighty community based participants with subjective memory complaints Community-based study at UCLA were randomized to one of three conditions: Memory Training, the program consisted of weekly 120-minute classes featuring instruction in three specific strategies: Method of Loci; Chunking Technique; and Face-Name Association, Health Education or Wait-List over seven weeks. All participants underwent pre- and 1-week post-intervention follow-up memory testing for recalling word lists (in serial order and any order) and face-name pairs. Memory control beliefs were assessed at baseline and follow-up using the Memory Controllability Inventory, which consists of four subscales; Present Ability; Potential Improvement; Effort Utility; and Inevitable Decrement. Results: Sixty-three participants (mean age [SD] 68.3 [6.7] years) were included in the analysis. ANCOVA revealed significant group differences in the Present Ability subscale, F2,58 = 4.93, p =.01. Participants in the Memory Training group significantly improved on the Present Ability subscale compared to the Health Education group (mean difference =.96, SE =.31, p =.003, effect size = 0.93). From regression analyses, baseline Memory Controllability Inventory subscales did not significantly predict memory performance after memory training. Conclusions: Baseline memory control beliefs did not predict memory performance following the intervention, but participating in memory training enhanced memory control beliefs about current memory function. These results suggest that participating in memory training can enhance confidence in one's memory ability.