Factors impacting on whether and how attitude predicts behavioral outcomes are crucial for attitude research. Previous theoretical and empirical attitude researches focused mainly on explicit attitude. According to implicit social cognition, individuals can hold simultaneously explicit and implicit evaluation toward the same object. Moreover, for some attitude object, explicit attitude assessed by traditional self-report measure is unrelated qualitatively to implicit attitude, which mostly based on the response-time or indirect measures such as IAT, GNAT or EAST. Associative-propositional evaluation model(APE, Gawronski & Bodenhausen, 2006, 2007), one of the most influential dual-process model in the implicit social cognition, provides the integrative theoretical framework for understanding the relationships between explicit attitude, implicit attitude and behavior(Hofmann, Gschwendner, Friese, Wiers, & Schmitt, 2008; Hofmann, Rauch, & Gawronski, 2007). Much less attention has been placed on the role of emotion in the relationship of explicit-implicit correspondence and implicit attitude-behavior. Some studies have revealed that emotional states influence the relationship between explicit attitude and implicit attitude. Compared with individuals with negatively feelings, individuals who experience a lot of positive emotions showed higher explicit-implicit attitude consistency(Huntsinger & Smith, 2009; Huntsinger, Sinclair, & Clore, 2009). An important but often ignored research topic, the role of emotional states in the relationship among explicit attitude, implicit attitude and behavior remains to be answered. Based on APE model, this study aimed to explore the role of emotion played in the transforming implicit attitude into explicit attitude, and the relationships between emotion, attitude and behavior accordingly. Subjects aged 19 through 22 years old were recruited on a university campus and paid 5 RMB for participating. The study consisted of two sessions. In the first session, 123 subjects completed a self-report questionnaire for assessing their explicit attitude toward eight kinds of food (including fruit and potato chips). A week later, 64 subjects agreed to participate a supposedly unrelated study and completed the Implicit Association Test (IAT) to assess their implicit attitude toward potato chips and PANAS. After that, an experimenter requested subjects to view randomly one of movie clips (comedy clips for positive emotion induction or tragedy clips for negative emotion induction) for the purpose of having a rest, and give them fruit and potato chips to eat 5 minutes before the end of movie clips. The weight of potato chips eaten in 5 minutes was measured and recorded as a dependent variable. As before, this study has verified the predictive effect of explicit attitude on the individuals’ behavior. The more positive attitude toward potato chips individuals hold, the more they eat during the viewing the movie. But this effect seemly cannot be applied to implicit attitude. Whether implicit attitude influences behavior relies on individuals’ emotional states, with implicit attitude influencing behavior among individuals in the state of positive emotional feeling, but not individuals with negative emotion. These results consist with the central hypothesis of Dual-Process Models of Social Information (Fazio, 1990; Strack & Deutsch, 2004; Hofmanna, Friese & Wiers, 2008) and Associative-Propositional Evaluation Model (Gawronski & Bodenhausen, 2006, 2007), and demonstrate the critical role emotion plays in implicit attitude-behavior relation. Besides, theoretical significance and limitation of this study and future directions were also discussed. Keywords: associative-propositional evaluation model; explicit attitude; implicit attitude; emotion