There is now a substantial literature investigating virtual reality exposure therapy (VRET) as a viable treatment option for anxiety disorders. In this meta-analysis we provide effect size estimates for virtual reality treatment in comparison to in vivo exposure and control conditions (waitlist, attention control, etc.). A comprehensive search of the literature identified 13 studies (n=397) that were included in the final analyses. Consistent with prediction the primary random effects analysis showed a large mean effect size for VRET compared to control conditions, Cohen's d=1.11 (S.E.=0.15, 95% CI: 0.82-1.39). This finding was consistent across secondary outcome categories as well (domain-specific, general subjective distress, cognition, behavior, and psychophysiology). Also as expected in vivo treatment was not significantly more effective than VRET. In fact, there was a small effect size favoring VRET over in vivo conditions, Cohen's d=0.35 (S.E.=0.15, 95% CI: 0.05-0.65). There was a trend for a dose-response relationship with more VRET sessions showing larger effects (p=0.06). Outcome was not related to publication year or sample size. Implications are discussed.