Previous studies showed attentional bias contains the rapid orientation of attention, the avoidance of attention and the difficulty to disengage from threatened stimuli. However, the time course of these tree components is still unclear. The present study is trying to explore the time-course of the threat-related attentional biases. The threatened and neutral emotional pictures were used as the stimuli. Subjects were asked to identify whether two pictures belonged to the same category (neutral or threatened) or not, in which two pictures were presented on computer screen simultaneously. Subjects’ eye movements were recorded during their task. Twenty six undergraduates volunteered to participate in the experiment. All stimulus- pictures were selected from the International Affective Picture System. Four types of stimuli pairs (i.e. threatened vs. threatened, threatened vs. neutral, neutral vs. threatened and neutral vs. neutral) were constructed from 152 pictures (77 neutral pictures vs. 75 threatened pictures). Three neutral pictures of stimuli pictures were repeated to pair twice and five threatened pictures of them were repeated to pair twice. Each type of stimulus pairs included 20 picture-pairs. Totally, there were 80 picture-pairs in all conditions. They were randomized and segregated into 4 blocks. Subjects were positioned in chin and forehead rest and seated 70cm in front of a 17-inch screen on which theses stimulus-pairs were present. Eye movements were recorded by SMI eye-tracker (iView Hi-Speed) with a sampling rate of 350Hz. In each trial, a white “+”was presented for 1000ms. Then a picture-pair was presented until 1000ms after subjects responded with maximum duration of 8000ms. The inter-trial interval (ITI) is 2000ms. The results showed that: (1) Subjects were more likely [F(1,25)=21.867, p<0.01] and faster (shorter saccadic latency) [F(1,25)=16.507, p<0.01] to move their eyes toward left visual field than that toward right, indicating strong attentional bias toward left. (2) The latency of the first saccade toward right visual field under “neutral vs. neutral” picture pair was significantly longer than that of “threatened vs. neutral” pair [F(1,25)=13.970, p<0.01]. Although the latencies of the first saccade toward left were faster than that of toward right under “threatened vs. threatened” [F(1,25)=7.690, p<0.01], “neutral vs. threatened ” [F(1,25)=15.462, p<0.01] and “neutral vs. neutral” [F(1,25)=20.483, p<0.01] conditions, it wasn’t significantly under condition of “threatened vs. neutral”. These results showed that subjects were reluctant to move their eyes to the threatened picture, suggesting that the avoidance of attention occurred less than 250ms, because all the first saccade latencies were shorter than 250ms. (3) In left visual field, the gaze duration under “threatened vs. neutral” was significantly longer than under “neutral vs. threatened” [F(1,25)=11.315, p<0.01]. In the right field, they were opposite [F(1,25)=6.777, p<0.01]. That was, subjects gazed longer, it was about 400ms to 600ms, at the threatened picture no matter where it was, and they need at least 600ms to move their eyes away. These results suggested that the difficulty to disengage from threatened picture. (4) The reaction time under “threatened vs. neutral” was the longest (200ms longer than other conditions) and the error rate was also the highest (5% higher than other conditions). It showed that subjects allocated more attention to left side of visual field. However, putting together with the attention avoidance and difficulty to disengagement made attention allocation more complicated. During comparing the two pictures at left and right visual field, the switching frequency was the lowest under “threatened vs. neutral” pair. It might increase the error rate of judgment indication that attentional bias influenced other cognitive process. Our results suggested that the attention avoidance and bias to left-side interacted in early stage, and the attention avoidance (less than 250ms) occurred earlier than the disengagement difficulty of attention (400ms-600ms).