Objective: As cognitive dissonance is theorized to contribute to the effects of dissonance-based eating disorder prevention programs, we evaluated a high-dissonance version of this program against a low-dissonance version and a wait-list control condition to provide an experimental test of the mechanism of intervention effects. Method: Female college students (N = 124, mean age = 20.9 years, SD = 3.9) with body image concerns were randomized to the 3 conditions. The high-dissonance program was designed to maximize dissonance induction, and the low-dissonance program was designed to minimize it; the substantive content of the 2 programs was matched. Results: Relative to controls, those in the high-dissonance condition showed significantly greater reductions in thin-ideal internalization, body dissatisfaction, dieting, and eating disorder symptoms by posttest, and those in the low-dissonance condition showed significantly greater reductions in the first 3 outcomes by posttest, with most of these effects persisting to 3-month follow-up. High-dissonance participants showed significantly greater reductions in eating disorder symptoms than low-dissonance participants did by posttest, but this effect was nonsignificant by 3-month follow-up. Conclusions: Results suggest that dissonance induction contributes to intervention effects but imply that the intervention content, nonspecific factors, and demand characteristics play a much more potent role in producing effects. (Contains 2 tables, 1 figure and 3 footnotes.)