In a 2 by 2 factorial design, high or low attraction for a simulated player (SP) was induced in 40 female college Ss. During a mixed-motive interaction, the SP sent intermittent promises of cooperation to Ss; the promises were either 10% or 90% credible. Results indicated a main effect of promise credibility on S cooperativeness on message-relevant trials. Liking for and evaluation of the SP were inversely related to perceived potency. Changes in attraction from pre-to postmeasures supported an expectancy theory of attraction. Schlenker et al (1971) found no effect of interpersonal attraction for a simulated source of contingent promises upon S compliance. Tedeschi (1973) has suggested that attraction elicits rewarding or cooperative behavior only when the source of benefits is perceived as acting in an altruistic and nonobligatory fashion. Contingent promises may be perceived as contractual agreements, a type of economic exchange (Blau, 1964) which renders liking irrelevant. A noncontingent promise, which offers unilateral benefits, should promote the effects of attraction on the influence process. The first hypothesis of the present study was, therefore, that Ss would comply with noncontingent promises more often under conditions of high attraction than low attraction (HI).It should be expected that the proportion of times a promisor does what he says he will do (Le., credibility) will affect a target's reactions. A replication of the Gahagan & Tedeschi (1968) finding that target Ss were more cooperative in a Prisoner's Dilemma Game (pDG) when the source of noncontingent promises was most credible was predicted (H2).There is evidence that liking for a person arouses the expectancy that the other will provide benefits, while disliking arouses the opposite expectancy (Kaufmann, 1967). It is plausible that under high attraction Ss set higher comparison levels (Le., expected gains from the interaction) than Ss under low attraction. These differences in expectancies would serve as a basis for disappointment or pleasant serprise and subsequent revision of liking for the other person (cf. Jones & Gerard, 1967; Thibaut & Kelley, 1959 have found support for the latter relationships. Furthermore, Ss who like a promisor should expect credibility to be high, since the congruence of verbal statement and behavior imply both the reliability of stated intentions and the beneficient cooperation of the promisor to achieve a joint gain. Conversely, low attraction should be associated with the expectancy that the promisor would not keep her word to be unilaterally benevolent and might in fact use false statements of intention to lure the S into an exploitative interchange. High attraction should lead to disappointment when the source's credibility is low, and low attraction should lead to pleasant surprise when the source's credibility is high. It is, therefore, predicted (H3) that in the high-attraction condition, high credibility will not change measured values of liking but low credibility will cause a decrement in liking, whi...