This study compares the predictive power of two models of moral judgment: (a) Kohlberg's model, which predicts stage consistency across varying content, and (b) a model that predicts variations in stage as a function of complex interactions between people and social contexts. Forty university students were asked to make and justify choices on three moral dilemmas—a dilemma involving the sale of defective merchandise (selling dilemma) and two dilemmas from Kohlberg's test of moral judgment. Half the participants made hypothetical choices on the selling dilemma; the other half were offered a selfish incentive—money—for concealing the defects in the merchandise. In support of the interactional model, (a) participants scored lower on the selling dilemma than on Kohlberg's test, (b) the monetary incentive affected moral choices, (c) participants justified the choice to conceal defects in the merchandise with significantly lower stage moral judgments than they invoked in support of the decision to disclose its defects, and (d) participants who upheld the choice to disclose supported it with higher stage justifications than those who rejected it. Surprisingly, however, participants who stood to make money disclosed more about the defects in the article and charged less than participants who did not.