Inhibition of return (IOR) forbids people’s attention to already attended objects, and promotes their detection to the new information in the scene during visual search. Previous studies showed that the magnitude of IOR decreaced when the face with some personal important social information appeared in the cueing location. An abundance of studies supported self-face was important to individuals and enjoyed the ability to withhold attention. However, whether self-face affects IOR was unknown.To meet the end, the impact of self-face on the IOR was tested. In addtion, there are several ways by which self-face could adjust the magnitude of IOR.The ex-Gaussian distribution analysis was also applied because it produced more directed and useful index to reveal how self-face affects IOR in detail. In the present study，the classic exogenous cue-target paradigm was applied to investigate the impact of different face types (self vs unfamiliar) on the effect of IOR as the SOA changes (350ms, 550ms, 750ms, 950ms, 1150ms ). A trail started with a drift-check, followed by the presentation of a white fixation with two position holders. Then, a cue face appeared on either the right or left position holder with equal probability for 250 ms, which was followed by the central cue. At last, the target was showed equally probably on the valid position (the same side of the cue present) or invalid position (the opposite side of the cue present) .The onset asynchrony of the cue and the target altered as the SOA was manipulated. In the study, the targets consisted of a line and a circle. Participants only responded to the line by pressing “Space” key while inhibiting themselves from making a response to the circle. They were instructed to respond to the target as accurately and quickly as possible. When no response was detected, the target would disappear in 1000ms. The ex-Gaussian distribution analysis was employed in the study to reveal how face influenced IOR in detail . 18 college right-handed students participated. Only reaction time of correct responses was calculated. Results showed that the reaction time to target in valid-cued locations was significantly longer than that in invalid-cued locations, indicating an IOR effect. Compared with the condition where other-face worked as a cue, self-face cue caused significantly less magnitude of IOR, showing that self-face mitigated the effect of IOR. However, no main effect of SOA or interaction of SOA with cue type was found on the magnitude of IOR. It suggested self face had no influence on the time course of IOR. In addition, the results of ex-Gaussian distribution analysis showed that the differences of response time between valid and invalid cued location were only reflected in the Gaussian component μ which is highly related to perception. By contrast, the decreased magnitude of IOR was caused by two self-face was reflected in the exponential component σ, which is related attentional decision. Our study showed self-face modulated the magnitude of IOR but not influenced its time course. And the result from ex-Gaussian distribution analysis provided much clearer evidence that the IOR may occur automatically and the influence of the self-face on the IOR was mainly due to that self-face faciliates particpants to make responses to the targets which appeared in cued place.