A meta-analysis of published studies in which scores on objective (i.e., self-report) or projective measures of interpersonal dependency were used to predict some aspect of dependency-related behavior revealed that validity coefficients for projective tests (number of comparisons = 32)were generally larger than validity coefficients for objective tests (number of comparisons = 54). The relationships of setting in which data were collected, sourer of behavioral ratings, and participant classification method on observed test score-behavior correlations were also assessed. Implications of these findings for use of objective and projective dependency measures in clinical, laboratory, and field settings are discussed.