We report a meta-analysis of research on the relationship between performance in extrasensory perception (ESP) tasks and the psychological trait extraversion. The meta-analysis comprises 60 independent studies, 17 independent investigators, and 2,963 subjects. The overall weighted mean correlation is small (r = .09), significant (z = 4.63, p = .000004), and nonhomogeneous. For forced-choice ESP studies (N = 45), the ESP/extraversion relationship appears to be an artifact of subjects' knowledge of their ESP performance upon their responses to the extraversion measure: evidence for the relationship is limited to studies where subjects completed the ESP task prior to extraversion assessment (N = 18 studies, r = .17, z = 3.51); no evidence for an ESP/extraversion relationship exists in studies where extraversion was assessed before the ESP task (N = 16 studies, r = -.02, z = -0.78). The two correlations differ significantly (z = 3.58, p = .00045). For free-response studies, a significant ESP/extraversion relationship exists that is free of this problem: extraversion testing preceded the ESP task in 11 of the 14 free-response studies (r = .21, z = 4.57, p = .000005). The ESP/extraversion relationship is both significant (r = .20, z = 4.46, p = .0000083) and homogeneous in the subset of free-response studies involving individual testing (N = 12 studies). The effect is homogeneous across investigators and extraversion scales. We also report a new confirmation of the ESP/extraversion relationship using the Extraversion/Introversion Scale of the Myers-Briggs Type Indicator. The effect size (r =. 18, t = 2.67, 219 df, p = .008) is very close to the meta-analytic estimate for free-response studies (r = .20) and is homogeneous across the eight experimenters. While the relationship between extraversion and ESP in the forced-choice studies is probably artifactual, we conclude that there is a significant ESP/extraversion relationship in the free-response studies, that the relationship is consistent across investigators and scales, and that meta-analysis of parapsychological research domains has predictive validity.