This paper begins by reviewing previous research concerning the external validity of mixed-motive games as models of international conflict, interpersonal behavior, and behavior in large-scale social dilemmas. Two further experiments are then described, both of which cast further doubt upon the usefulness of such games as models of any real-world reference system. The author cites the large number of publications existing in this area, and compares it to the small number of validation attempts, suggesting that such studies are worth carrying out only if a correspondence can be established between behavior in the game nad behavior in an external reference system. Because of the difficulty of establishing such a correspondence, the author suggests redirecting validation attempts away from individual analysis, to focus more on situational variables in games and real-world dilemmas. (Author/BP)