Background: Virtual reality (VR) endoscopy simulators are increasingly being used in the training of novice endoscopists. There are, however, insufficient data regarding the effect of simulator training on the early learning curve of novice endoscopists. Objective: The aim of this study was to assess the clinical performance of novice endoscopists during colonoscopy after intensive and prolonged training on a VR endoscopy simulator. Design: Prospective study. Setting: Single university medical center. Patients: Patient-based assessment (PBA) of performance was carried out on patients routinely scheduled for colonoscopy. Interventions: Eighteen trainees without any endoscopic experience were included in the study. They were divided into 2 groups. The simulator-training program consisted of either 50 (group I) or 100 (group II) VR colonoscopies. After 10, 30, and 50 (group I) and after 20, 60, and 100 (group II) VR colonoscopies, trainees underwent both simulator-based assessment and PBA. Main Outcome Measurements: Cecal intubation time, colonic insertion depth, and cecal intubation rate. Results: Eighteen novices participated in the study. All completed VR training and assessments. The mean cecal intubation time on the SBA decreased from a baseline of 9.50 minutes to 2.20 minutes at completion of the training (P = .002). Colonic insertion depth during PBA improved from 29.4 cm to 63.7 cm (P < .001). The learning effect of simulator training ceased after 60 colonoscopies. Limitations: Single-center study, no formal sample size calculation. Conclusions: VR training by using a colonoscopy simulator leads to a significant improvement in performance with the simulator itself and, more importantly, to significantly improved performances during patient-based colonoscopy. This study demonstrates the rationale for intensive simulator training in the early learning curve of novices performing colonoscopy.