Objective As the fiberoptic gastroscopy using midazolam is being in widespread use, the exact nature of midazolam on memory should be clarified. We intended to examine whether midazolam causes selective anterograde amnesia and what impact it has on other aspects of memory and general cognitive function. Methods We recruited healthy subjects undergoing fiberoptic gastroscopy under conscious sedation. At baseline, history taking for retrograde amnesia and the Korean version of the Montreal Cognitive Assessment were performed. A man's name and address were given immediately after intravenous midazolam administration. After gastroscopy, the subjects were asked to recall those items. By the time they had fully recovered consciousness, the same test was repeated along with the Korean version of the Montreal Cognitive Assessment and a test for retrograde amnesia. Results A total of 30 subjects were enrolled in this study. Subjects with high-dose midazolam showed lower scores in the immediate and delayed recall of "a man's name and address" compared with those with low-dose midazolam. The midazolam dose was inversely correlated with the delayed recall scores of "a man's name and address." On full recovery of consciousness, the subjects did not exhibit any of anterograde or retrograde amnesia. Conclusions These findings suggest that midazolam causes transient selective anterograde amnesia in a dose-dependent manner.