Objective To explore the value of eye movement analysis in screening mild cognitive impairment（ MCI）. Methods Databases like Pubmed， Embase， Cochrane Library， EBSCO， OVID， ChineseBiomedical Literature（ CBM），CNKI， Wanfang and VIP were systemically searched for diagnostic trial of eye-tracking in screening patients with mild cognitive impairment published before March 2020. The NOS quality evaluation tool was used to evaluate the quality of the literatures that met the inclusion criteria. The relevant data of research characteristics and eye movement indexes（ saccade reaction time and saccade error rate） were extracted and analyzed by RevMan5.3 Meta analysis software. Results Finally， 6diagnostic trials involving 514 cases were included， among which 219 were diagnosed as MCI by Petersen creteria. The MD and 95%CI of antisaccade（ gap paradigm） reaction time（ AST gap Rt） in 6 studies was 23.37，（4.51-42.22）. The MD and 95%CI of antisaccade（ gap paradigm） error rate in 4 studies was 15.50，（14.99-16.01）. The difference between the MCI group and the control group was statistically significant（ P＜ 0.001）. The MD and 95%CI of prosaccade（ gap paradigm） reaction time in 2 studies was 13.68，（-8.88-36.25）， with no statistical difference between the two groups（ P=0.230）. Conclusions Theeye movement indexes， such as antisaccade reaction time and antisaccade error rate， can be used to distinguish MCI patients and normal subjects，while the prosaccade reaction time has little significance for screening MCI patients.