Two studies examined the causal role of emotional arousal in self-evaluation maintenance processes. In previous work, Tesser and Campbell (1982) found that Ss were most charitable in their perception of another's performance when self-relevance was low and the other was close. If emotional arousal mediated this pattern of behavior, then the pattern of behavior should be replicated when arousal is present but attenuated when arousal is misattributed or low. The misattribution hypothesis was tested in Study 1. Study 2 was a correlational study in which physiological arousal and misattribution were measured. The results supported the prediction that SEM processes would be attenuated when arousal was attributed to external sources. The misattribution effect was particularly pronounced among more highly aroused Ss. The results are discussed in terms of their implications for the SEM model and person perception.