A study tested two hypotheses concerning credibility (an auditor's attitude toward a source) and arousal (attention level of an auditor): (1) the effect on learning of credibility will interact with the effect of the vigilance level of auditors, and (2) under conditions of low arousal, auditors will learn significantly more from a message if it is attributed to a higher credibility source than if it is attributed to a lower credibility source. A single, trained female student delivered the same informative message on the effects of TV violence to 144 undergraduates enrolled in eight sections of a basic university speech course. Her introduction and the prospect of a test were used to produce higher and lower levels of credibility and vigilance. Results of immediate and delayed posttests confirmed both hypotheses: the effect of credibility on learning did interact significantly with the vigilance level of the learning situation, and higher credibility produced significantly more right answers than lower credibility in the low vigilance condition but not in the high vigilance condition. In an unpredicted result, women scored higher than men, a fact perhaps explainable by uncontrolled differences in intelligence or interest. (JL)