Background/Aims: The necessity for pharyngeal anesthesia during upper gastrointestinal endoscopy is controversial. This study aimed to compare the observation ability with and without pharyngeal anesthesia under midazolam sedation. Methods: This prospective, single-blinded, randomized study included 500 patients who underwent transoral upper gastrointestinal endoscopy under intravenous midazolam sedation. Patients were randomly allocated to pharyngeal anesthesia: PA+ or PA– groups (250 patients/group). The endoscopists obtained 10 images of the oropharynx and hypopharynx. The primary outcome was the non-inferiority of the PA– group in terms of the pharyngeal observation success rate. Results: The pharyngeal observation success rates in the pharyngeal anesthesia with and without (PA+ and PA–) groups were 84.0% and 72.0%, respectively. The PA– group was inferior (p=0.707, non-inferiority) to the PA+ group in terms of observable parts (8.33 vs. 8.86, p=0.006), time (67.2 vs. 58.2 seconds, p=0.001), and pain (1.21±2.37 vs. 0.68±1.78, p=0.004, 0–10 point visual analog scale). Suitable quality images of the posterior wall of the oropharynx, vocal fold, and pyriform sinus were inferior in the PA– group. Subgroup analysis showed a higher sedation level (Ramsay score ≥5) with almost no differences in the pharyngeal observation success rate between the groups. Conclusions Non-pharyngeal anesthesia showed no non-inferiority in pharyngeal observation ability. Pharyngeal anesthesia may improve pharyngeal observation ability in the hypopharynx and reduce pain. However, deeper anesthesia may reduce this difference.