Drawing upon the multiple roles of affect in the ElaborationLikelihood Model proposed by Petty， Cacioppo， and their colleagues (Petty， Cacioppo， & Kasmer， 1987; Petty， Cacioppo， Sedikides， & Strathman， 1988; Petty， DeSteno， & Rucker， 2001; Petty， Gleicher， & Baker， 1991)， this study explored the different roles that affect plays in the ad message reception process for a high involving and a low involving product. As expected， findings showed that， when processing ad messages for a high involving product， participants in positive affective states responded more favorably to a positive ad cue than to a negative cue when ad arguments were weak， whereas ad cue valence did not influence their responses when ad arguments were strong. On the contrary， participants in negative affective states responded more favorably to ads with strong arguments than weak arguments but did not respond differently to ads with positive versus negative cues. However， the proposed main effect of affective state on ad judgments for a low involving product did not emerge.