It was hypothesized that source esteem would be effective in mediating target compliance when the source uses threats as an influence tactic, particularly if esteemed individuals maintain a facade of truthfulness, because threat credibility would then be overestimated. Threats from a low esteemed source would have lower credibility and lead to infrequent compliance. The experimental paradigm used a message-modified version of the Prisoner's Dilemma Game (PDG) in which a simulated source could send explicit contingent threats to target subjects during the course of a mixed-motive conflict situation. An experimental confederate first carried out a rehearsed esteem inducement procedure. The game was then played. In the experimental group, 20 threats were sent (as an illuminated light) over 150 game trials, and half the subjects were punished for non-compliance. Subjects then rated the confederate. The subjects were found to comply more frequently in high source esteem conditions irrespective of threat credibility. In low source esteem conditions, subjects complied as a direct function of threat credibility. (KS)