We performed a randomized prospective trial to compare unsedated endoscopy via transnasal and transoral routes using a small-caliber endoscope. Two hundred patients referred for diagnostic esophagogastroduodenoscopy (EGD) were randomly allocated to two groups: those undergoing transnasal (TN) and transoral (TO) endoscopy. We examined the insertion rate, examination duration, nasal pain, pharyngeal pain, number of occurrences of pharyngeal reflex, severity of discomfort throughout the examination, and rate of adverse events. Patients were asked to rate the severity of their pain or discomfort on a 10-cm visual analog scale (VAS). We identified statistically significant differences between the TN and TO groups in rate of insertion (95% versus 100%, respectively, P < 0.05) and examination duration (9.7 +/- 3.0 min versus 8.4 +/- 2.7 min, respectively, P < 0.005). Severity of discomfort throughout the examination was comparable in the TN and TO groups (3.0 +/- 1.8 versus 2.9 +/- 2.3, NS). Nasal bleeding occurred as an adverse event in 4.1% of patients in the TN group. Patients in the TO group were more likely than those in the TN group to prefer the present method in the subsequent endoscopic examination (99.0% versus 82.1%, P < 0.00005). These results indicated that transoral insertion is superior to transnasal insertion in endoscopy procedures performed with small-caliber endoscope.