As an important cue of social cognition, faces usually bring lots of information about target’s characters, affecting a variety of domains (Zebrowitz, 2006). therefore, research on facial cognition has gradually been extended to traits inference based faces, such as the facial cues used to inference traits, the characteristics which can be inferred and so on. with the study development, more and more researchers pay attention the accuracy of facial traits inference. a series of studies have been carried out. Firstly, this paper introduce the concept of Face-traits Inference and the relationship between faces and traits. Face-traits Inference is a process of observers inferring other’s traits from faces. On the one hand, observers can read traits from faces. On the other hand, observers can read traits into faces. Then, we account for its theoretical basis from Physiology, evolution and cognitive orientation. Changes in facial structures caused by hormones and maturity can explain this phenomenon from physiology, but it cannot explain the inference of traits related to social relationship. The two main theories in cognitive psychology: stereotypes and overgeneralization effects establish a connection between faces and traits, but these theories do not explain the mechanisms behind the connection. Facial traits inference may be the product of evolution, because some researchers found that the ability to inference traits from faces formed in the early age, with the character of automated processing (Hassin & Trope, 2000; Cogsdill et al. 2014). Due to the important meaning of face-traits inference, researchers apply objective index and subjective index to measure the accuracy of face-traits inference, the objective index including the consistency between the judges, consistency between self and others and the prediction power of behavior. The mainly used subjective index is confidence score. means the confidence of inferring traits from faces, Based on the above accuracy index, It was found that trait inference from faces has moderate accuracy (Wolffhechel et al., 2014). The accuracy of facial trait inference is mainly influenced by three factors of observers, targets and situations. The physiologic factor of targets, such as age, sexual and facial attractiveness affect the accuracy. Interestingly, attraction has two opposite effects on the accuracy. Characters of targets, such as the physiologic factors, mental conditions, experience and cultural background are also important factors affect facial traits inference. For example, needs and motivation of observers influence facial trait inference, it was shown that individuals who are eager to establish intimate relationships with others, are more accurate in detecting affinity cues from faces (Bernstein et al., 2008). Situations and task are also important factors influence the accuracy of trait inference. Finally, the current research proposed some questions that still need further clarification. First of all, previous studies have neglected the interaction between facial configuration and expressions. secondly, previous studies lacked integration of accuracy indicators. thirdly, the recent research focus on the various factors that affect the accuracy of facial trait inference. However, there is no comprehensive consideration of the interaction of these factors. Finally, the theory of facial trait inference needs to be integrated. Regarding the issue above, future research should be carried out from the following aspects. Firstly, in terms of research methods, we should enrich the accuracy index. Secondly, the influence factors of facial trait inference should be comprehensively investigated. Thirdly, clarify the facial cues of traits inference and construct an accurate face-traits inference model. Last but not the least, researchers need to combine the three orientations of physiology, evolution and cognition, at the same time, focus on the interaction between observers and targets in specific situations, so that understand the mechanism under this process and construct a systematic theoretical model.