Trait anxiety refers to individuals’ relatively stable tendency to anxiety and is an anxiogenic reaction to environmental threats. Studies have found that individuals with high trait anxiety have abnormal attention bias. In previous studies, the stimulus presentation time was set in advance, and the response time or accuracy of the subjects were recorded. The investigation of attention characteristics may have indirect speculation, and it is difficult to reflect the dynamic changes of attention characteristics of the same stimulus. In addition, few focus is given to stimuli other than threatening stimuli. In this experiment, the eye movement characteristics of adolescents with trait anxiety tendencies in free viewing of emotional face images were used as indicators to examine their attention characteristics when processing emotional stimuli. This experiment is a two-factor mixed experiment design. The independent variables were interpersonal factor trait anxiety (high trait anxiety vs. low trait anxiety) and internal factor emotional properties (happy vs. fear vs. anger vs. neutral). The dependent variable was the subject’s eye movement index, including the time to first fixation (TFF), the first fixation duration (TFD), the fixation counts (FC), and the average fixation duration (AFD) of the subjects in the area of interest (AOI). The results showed that subjects in the high anxiety group had faster arrived time on anger and fear faces for first fixation and the shorter first fixation duration, while slower arrived time on happy faces for first fixation and the shorter first fixation duration. After 1000ms of stimulation presenting, they stared more at the anger and fear faces, the average duration of fixation was shorter. While the fixation number on happy face was smaller, and the average duration of fixation was longer. In the low trait anxiety group, although the arrival time of the first fixation of anger and fear faces was earlier than that of calm faces, the difference between them was not significant; the first time to happy faces arrived faster and the first fixation lasted longer. They had fewer fixation counts on anger and fear faces, the average duration of fixation was longer; the counts of fixation on happy faces was more, and the average fixation duration was shorter. This indicates that in the early stage of cognitive processing, adolescents with high trait anxiety tend to threatening stimuli existed with attention-oriented facilitated. They do not have difficulties in attention disengagement. The processing mode presented a phenomenon of “slightly tasted”. Also they do have attention avoidance to positive stimuli. Adolescents with low trait anxiety tend not to show obvious attention bias to threatening stimuli, and attention characteristics to positive stimuli are biased in the early stage. In the late stage of cognitive processing, adolescents with high trait anxiety tend to “ruminate” in cognitive processing on threat stimuli, and the attention characteristics of positive stimuli still exist. Adolescents with low trait anxiety tend to avoid attention to negative stimuli, and maintain attention to positive stimuli.