Study Objective: To investigate the interaction of midazolam with different doses of bupivacaine, by comparing the analgesic, sedative, and amnesic effects of continuous epidural midazolam with two different doses of bupivacaine. Design: Prospective, randomized study. Setting: Operating room and intensive care units of a university hospital. Patients: 100 ASA physical status I and II postgastrectomy patients (40–70 yrs) without any complications. Interventions: Patients were divided into four groups (n = 25) and administered continuous thoracic epidural infusion (40 mL/12 hr) of the drugs via a balloon infuser. The contents of the infuser (40 mL) were bupivacaine 180 mg with midazolam 20 mg (HM group), 90 mg with midazolam 20 mg (LM group), 180 mg without midazolam (HC group), or 90 mg without midazolam (LC group). As a rescue medication, 50 mg indomethacin suppository was the first choice, then IM pentazocine 15 mg. Measurements: Analgesia and sedation scores, blood pressure (BP), heart rate, respiratory rate, oxygen saturation, amnesia, the number of rescue medications, and time to the first rescue medication were monitored. The area under the curves (AUCs) of analgesia and sedation scores, and BP were calculated. Main Results: In the HM group, significantly better analgesia and sedation were obtained and the number of rescue medications given was the lowest of the four groups. The time to the first rescue medication was longest in the HM group, followed by the LM group, then the HC, and, finally, LC groups. The numbers of the patients with amnesia were greater in the HM and LM groups than the HC and LC groups. Blood pressure decreased significantly in the HM group but no treatment was necessary. The AUCs of analgesia and sedation scores, and BP were lowest in the HM group. Conclusions: Adding midazolam increased not only analgesic but also sedative effect with increasing dose of bupivacaine in a postoperative continuous epidural administration.