Research on Eye Movement Desensitization and Reprocessing therapy (EMDR) was reviewed to answer the questions "Does EMDR work?" and "If so, Why?" This first question was further subdivided on the basis of the control group: (a) no-treatment (or wait list control), (b) nonvalidated treatments, and (c) other validated treatments. The evidence supports the following general conclusions: First, EMDR appears to be effective in reducing at least some indices of distress relative to no-treatment in a number of anxiety conditions, including posttraumatic stress disorder, panic disorder, and public-speaking anxiety. Second, EMDR appears at least as effective or more effective than several nonvalidated treatments (e.g., relaxation, active listening) for posttraumatic stress reactions. Third, despite statements implying the contrary, no previously published study has directly compared EMDR with an independently validated treatment for posttraumatic stress disorder (e.g., therapist-directed flooding). In the treatment of simple phobia, participant modeling has been found to be more effective than EMDR. Fourth, our review of dismantling studies reveals there is no convincing evidence that eye movements significantly contribute to treatment outcome. Recommendations regarding further research directions are provided.