BACKGROUND: Despite colonoscopy represents the conventional diagnostic tool for colorectal pathology, its undeniable discomfort reduces compliance to screening programmes. AIMS: To evaluate feasibility and accuracy of a novel robotically-driven magnetic capsule for colonoscopy as compared to the traditional technique. METHODS: Eleven experts and eleven trainees performed complete colonoscopy by robotic magnetic capsule and by conventional colonoscope in a phantom ex vivo model (artificially clean swine bowel). Feasibility, overall accuracy to detect installed pins, procedure elapsed time and intuitiveness were measured for both techniques in both operator groups. RESULTS: Complete colonoscopy was feasible in all cases with both techniques. Overall 544/672 pins (80.9%) were detected by experimental capsule procedure, while 591/689 pins (85.8%) were detected within conventional colonoscopy procedure (P=ns), thus establishing non-inferiority. With the experimental capsule procedure, experts detected 74.2% of pins vs. 87.6% detected by trainees (P<0.0001). Overall time to complete colon inspection by robotic capsule was significantly higher than by conventional colonoscopy (556±188s vs. 194±158s, respectively; P=0.0001). CONCLUSION: With the limitations represented by an ex vivo setting (artificially clean swine bowel and the absence of peristalsis), colonoscopy by this novel robotically-driven capsule resulted feasible and showed adequate accuracy compared to conventional colonoscopy.