Social anxiety has been associated with abnormalities in cognitive processing in the literature, manifesting as various cognitive biases. To what extent these biases interrupt social interactions remains largely unclear. This study used the Social Judgment Paradigm that could separate the expectation and experience stages of social feedback processing. Event-related potentials (ERPs) in these two stages were recorded to detect the effect of social anxiety that might not be reflected by behavioral data. Participants were divided into two groups according to their social anxiety level. Participants in the high social anxiety (HSA) group were more likely to predict that they would be socially rejected by peers than did their low social anxiety (LSA) counterparts (i.e. the control group). Compared to the ERP data of the LSA group, the HSA group showed: (a) a larger P1 component to social cues (peer faces) prior to social feedback presentation, possibly indicating an attention bias; (b) a difference in feedback-related negativity amplitude between unexpected social acceptance and unexpected social rejection, possibly indicating an expectancy bias; and (c) a diminished sensitivity of the P3 amplitude to social feedback valence (be accepted/be rejected), possibly indicating an experience bias. These results could help understand the cognitive mechanisms that comprise and maintain social anxiety.