Objective To study the anterograde amnesia caused by midazolam which was used by patient-controlled sedation and anesthetist-controlled sedation.Methods 120 patients in surgical patients were divided ran-domly into two groups,60 patients in each.P group:patient-controlled sedation with midazolam;A group:anesthetist-controlled sedation with midazolam.Atropine 0.5mg was injected 30rain before operation,explaining the correct appli-cation of PCA pump to patients when we interview them.HR,ECG,BP,SPO2 were monitored during anesthesia.P group:midazolam 20mg+normal saline(NS) to 100mL,midazolam loding dose 0.02mg.kg-1,continous rates 0.001mg.kg-1.h-1,PCS 0.02mg.kg-1,lockout time 3min,PCS rates 30 seconds.A group:midazolam loding dose 0.02 mL.kg-1,continue rate 0.001mg.kg-1.h-1,Ramsay sedation score was used to evaluate the sedative depth every 10 minutes,appropriate sedative depth(Ramsay sedation score 4) kept by adjusting the infusion rate.Showing patients three pictures before sedation started,and checking patient memory at 12h after operation.Accessing the patient’s satisfaction degree by VAS score.Results The amount of midazolam was a little higher in P group(10±4)mg,than that in A group(9±3) mg,but there was no significant difference(P0.05).Postoperative anterograde amnesia existed in both two groups,memorability has no significant difference between two groups(P0.05),and there was a statistic difference between two groups [P group(89 ± 6) and A group(71 ± 8)] in patient satisfaction degree(P0.001).Conclusion Satisfactory anterograde amnesia occurred in both PCS group and ACS group.Patient preferred PCS.