Individuals with high social anxiety (HSA) show abnormal processing of emotional faces, which may increase their social anxiety. A growing number of event-related potential (ERP) studies have explored the neural mechanisms underlying the static-emotional face processing of HSA individuals. In view of the ecological validity of dynamic faces, this study will further explore the time course of dynamic-emotional face processing in individuals with HSA. To this end, 30 high and 30 low social anxiety (LSA) participants were asked to perform an identification task of dynamic-emotional faces while their brain responses were recorded using an ERP technique. The behavioral results showed the recognition accuracy of dynamic faces was higher than static faces when these faces were happy. For the P100 component, HSA participants showed higher P100 mean amplitudes of dynamic than static faces in the left hemisphere when they viewed happy, but not angry faces. In addition, increased N170 mean amplitudes of dynamic-happy faces were showed. Furthermore, the LPP mean amplitudes of dynamic faces were smaller than those of static faces. In sum, this study could provide a better understanding of the time course of dynamic-emotional face processing in HSA individuals.