Previous research suggested that high social anxiety individuals may form biased interpretations of positive facial stimuli. It was suggested that social anxiety individuals may discount the positive social information, or even interpret it in a negative fashion. This biased interpretation processing might tremendously influence social anxiety’s positive affect in a negative way, as well as their quality of life. So it’s important to learn more about social anxiety’s cognitive processing of positive social information. Recently, some researchers focused on how social anxiety individuals processed positive social information, however, most of these studies used self-report as their experimental method, and may suffer from several methodological limitations. The present study aimed to figure out whether social anxiety is characterized by an interpretative bias towards positive facial stimuli through a more objective and accurate method. The affective priming paradigm developed by Fazio, Sanbonmatsu, Powel, & Kardes (1986) was applied as a measure of socially anxious individuals’ biased interpretative pattern towards positive faces. And in the present study, the emotional faces were used as the primes, while the positive and negative words were used as targets. The current study consisted of a 2(prime valence: positive vs. scrambled faces) × 2(target valence: positive vs. negative words) × 2(group type: high vs. low social anxiety group) mixed factors design. Only the group type is a between-subjects factor. In each trial, a fixation was first presented for 500ms, followed by a positive or a scrambled face for 200ms. After the face disappeared, there was an interval of 50ms during which a grey screen appeared before a positive or a negative word was presented. This word remained on screen for 2000ms or until the participants pressed the reaction key. There was 1000ms interval between each trial. Participants were asked to evaluatively categorize the affectively polarized non-social word targets. And they were instructed to focus on the task, and try to respond to the target words as quickly and as accurately as possible. Their reaction time and correctness were recorded, while only the reaction time was analyzed with repeated measure ANOVA. After the prime task, participants were asked to evaluate the attractiveness of 10 neutral faces which were used to make the scrambled faces. Then, they were going to finish the Social Phobia Scale, and a questionnaire for personal information. After this, the participants were thanked and debriefed. The study found that the high social anxiety group and the control group exhibited different affective priming effects when primed by positive faces: In high social anxiety group, no affective priming was found; while the control group showed the typical congruency effect for positive facial primes. They reacted significantly more quickly to the positive targets than those negative words. In addition, the social anxiety group took significantly longer to respond to the positive target words than the controls. The results might suggest that social anxious individuals had an interpretation bias towards positive faces. They may make less positive interpretations on positive social information. At the same time, the study showed that the affective priming paradigm can be successfully applied to test the positive cognitive bias in social anxiety group.