BACKGROUND: Although midazolam administration may occasionally induce a paradoxical episode, such as threatened crying and violent behavior in children, systematic studies on the causes of paradoxical reaction are limited. We investigated the effect of children's age and a dose of midazolam on the paradoxical reaction. METHODS: A total of one hundred sixty four children of 1-3 years and 3-5 years, were enrolled in this study. Each age group randomly received 0.05 mg/kg or 0.1 mg/kg of intravenous midazolam (41 patients/group). RESULTS: The incidence of paradoxical midazolam reaction in the study groups, 1-3 years with 0.1 mg/kg of intravenous midazolam, 1-3 years with 0.05 mg/kg, 3-5 years with 0.1 mg/kg, and 3-5 years with 0.05 mg/kg were as follows: 29.3%, 12.2%, 7.3% and 2.4%, respectively. The incidence among the 4 groups was significantly different (P = 0.002), highest in the 1-3 years receiving 0.1 mg/kg of midazolam (29.3%). Both age (P = 0.004, OR [95%CI] = 5.3 [1.7-16.8]) and dose of midazolam (P = 0.036, OR [95%CI] = 3.0 [1.1-8.4]) were risk factors. Perioperative clinical data including anxiety scales of children were not associated with the paradoxical midazolam reaction. CONCLUSIONS: In conclusion, we suggest that children less than 3 years old receiving higher dose of intravenous midazolam are at risk for the paradoxical midazolam reaction.