Background/Aims: Modified neuroleptanalgesia (m-NLA) with midazolam is often used for sedation and analgesia during endoscopic submucosal dissection (ESD) for gastrointestinal neoplasia. However, interruption due to poor response to midazolam is often experienced during ESD for esophageal squamous cell carcinoma (ESCC) because most patients with ESCC have a history of heavy alcohol intake. We examined the incidence and risk factors for poor response to m-NLA with midazolam and pethidine hydrochloride. Methods: This retrospective cross-sectional study was conducted at a single institution. Between April 2007 and July 2013, 151 patients with superficial ESCC who underwent ESD under sedation using m-NLA with midazolam and pethidine hydrochloride were enrolled. Poor response to sedation was defined as the use of a second drug when Ramsay Sedation Score 1-2. Results: Poor response to sedation occurred in 66.2% patients. Most cases of poor response were controlled by using additional flunitrazepam. Multivariate logistic regression analysis showed that cumulative alcohol intake and major specimen size were independent risk factors for poor response to sedation (OR 3.63, 95% CI 1.20-10.99, and OR 3.23, 95% CI 1.26-8.25). Conclusion: Our study indicated that cumulative alcohol intake and major specimen size were associated with poor response to m-NLA with midazolam and pethidine hydrochloride.