The attention processing of emotional stimuli includes two stages: recognition and disengagement. Rapid recognition of emotional stimuli can ensure the high-efficiency processing of emotional information. However, if attention bias is too strong to be effectively removed, it may cause maladaptation to the environment and mood disorders. Therefore, the characteristics of attention to emotional stimuli in both the recognition and the disengagement stage must be investigated. The current study included two experiments. In experiment 1, we adopted a two-choice oddball task and used neutral faces as standard stimuli; crying and smiling faces were used as deviant stimuli. An EyeLinkⅡ desktop eye tracker with a 250 Hz sample rate was used to record the participants’ eye movements. In experiment 2, we used a cue-target paradigm to imitate the process of disengagement from emotional stimuli. We classified crying, smiling and neutral faces as the cue stimuli and asked the participants to respond according to the position of the target stimulus. We hypothesized that crying faces would be recognized with a shorter reaction time and higher accuracy than smiling faces in the recognition stage. Although the mouth was important for the detection of smiling faces, the eyes were more relevant for crying faces. In the disengagement stage, the types of facial expressions could affect the inhibition of return. Under valid cue conditions, processing of the target stimulus was faster after crying facial expressions were presented than after other facial expressions were presented. The results of experiment 1 suggested that compared to smiling faces, crying faces were recognized faster and with higher accuracy. In addition, the eye movement data indicated that whether shown a crying face or a smiling face, the participants’ fixation time, fixation time ratio and fixation count ratio were significantly greater in the eyes and nose area than in the mouth area. Furthermore, there was an interaction effect between facial expression types and AOIs (areas of interest). The fixation time and fixation time ratio were higher in the eyes area than in the mouth area when recognizing crying faces, whereas the fixation indexes were higher in the mouth area than in the eyes area when recognizing smiling faces. Experiment 2 showed that there was an IOR, and in accordance with the hypothesis, there was an interaction effect between cue validity and facial expression type. Under valid cue conditions, the average fixation time and saccade latency for the target stimuli after a crying facial expression was presented were significantly shorter than those after other facial expressions were presented. In conclusion, there are different attention biases to the crying facial expression in recognition and disengagement. In the recognition stage, individuals can recognize crying facial expressions accurately and react quickly. In addition, the eye-movement patterns of crying faces and smiling faces have both similarities and subtle differences. In the disengagement stage, crying facial expressions can facilitate the orienting and visual processing of target stimuli under valid cue conditions.