The human brain responds before the application of aversive stimuli. Anticipation allows the organism to prepare its nociceptive system to respond adequately to the subsequent stimulus. The uncertainty of occurrence and intensity of an aversive event may lead to increased anticipatory anxiety, fear, physiological arousal and sensory perception. How does the anticipatory anxiety induced by the intensity of an uncertain aversive event influence the processing of social stimuli such as other’s facial expressions? The present study used a revised version of the “monetary wheel of fortune” game to induce different levels of anticipatory anxiety. Two kinds of bets are set: 1 and 100, “1” means that the participants would win/lose 1 points (equivalent to 0.1 Yuan); “100” means that the participants would win/lose 100 points (equivalent to 10 Yuan). After the bet was chosen and before the outcome appeared, the participants were presented with pictures of facial expressions (including three kinds: happy, neutral and fearful). Participants were instructed to judge the valence of the expression by pressing buttons. The EEGs during the participants observing different facial expression under different levels of anticipatory anxiety were recorded and compared. Since this is a novel paradigm, we run a validation test with 20 participants who did not participate in the EEG studies. They were instructed to rate the anxiety level according to their current feelings after the fortune is done (1 or 100 was chosen) on a 9 point scale (from 1: not anxious at all to 9: extremely anxious). Results of the validation test indicated that 100 elicited a significantly higher level of anxiety than 1. Therefore, this novel paradigm could elicit a different level of anticipatory anxiety as we expected. On the behavioral level, we found that the participants showed a significant negative bias in which they were tending to under-rate the pleasantness of neutral face under the high anticipatory anxiety conditions. On the ERP level, a significant interaction of anticipatory anxiety and facial expressions was observed on the N170 component. Both happy and fearful face induced significantly greater N170 than neutral face induced N170 only in high anticipatory anxiety conditions but not in low anticipatory anxiety conditions. Crucially, we found that the fearful-face-induced N170 amplitude was significantly greater under high anticipatory anxiety condition than under low anticipatory anxiety condition, while the difference of N170 induced by happy and neutral faces between the two anticipatory anxiety levels was insignificant. In addition, we also found that that the modulation effect of anticipatory anxiety on the fearful-face-induced N170 is related to the personality trait scores of neuroticism and state-trait anxiety. Combined with behavioral and ERP findings, these results show that the level of anticipatory anxiety have an impact on the processing of facial expressions, and the processing of different expressions is affected by the anticipatory anxiety differently. Previous studies have shown that patients with various anxiety disorders, including phobia, generalized anxiety disorder and social phobia, may exhibit abnormal anticipatory anxiety. The present study focusing on the healthy population will help deepen our understanding of the mechanism of abnormality in the disease population.