Introduction: Neoplastic tissue and normal tissue are known to show different autofluorescence, when irradiated by blue light. Autofluorescence imaging system using fiberoptic endoscopy was developed and is under clinical use for the detection of gastrointestinal neoplastic lesions. Video endoscopy has better quality in the resolution of image compared with the fiberscopy. An Auto-Fluorescence Imaging (AFI) video colonoscope system (Olympus Optical Co., Ltd.) was already developed and we reported the clinical usefulness for detecting colonic neoplasms (DDW2003). Recently, prototype of AFI video endoscopy for upper GI tract was developed and we have evaluated this system for the diagnosis of gastric neoplastic lesions. Methods: The AFI video endoscopy for upper GI tract has two CCDs in the end of the scope for the white light conventional observation and the autofluorescence observation respectively. The white light conventional image is equivalent to that of a conventional video esophagogastroduodenoscope (XGIF-Q240, Olympus). An autofluorescence image (excitation: 395 - 475nm, detection: 490 - 625nm) and reflected green and red images (G: 540 - 560nm, R: 600 - 620nm) are captured with sequential method, and these three images compose a single pseudo-color AFI image on the monitor in real time. We performed the examination using AFI videoscopy under informed consent for 28 patients with 34 lesions; 11 adenocarcinomas, 5 adenomas and 18 non-neoplastic lesions (hyperplastic polyps, gastric ulcers and gastritis), which were histologically confirmed by the resected specimen. Results: Most of neoplastic lesions (cancers and adenomas) were displayed as magenta color in the background green color on the monitor. 10 of 11 cancers were correctly diagnosed by AFI video endoscopy, while one cancer was diagnosed as a benign ulcer. All 5 adenomas were correctly diagnosed by AFI video endoscopy. 16 of 18 non-neoplastic lesions were correctly diagnosed by AFI video endoscopy, while one lesion was diagnosed as cancer and one lesion was diagnosed as adenoma. Conclusions: Newly developed autofluorescence imaging videoscopy system for upper GI tract has ability for the detection of gastric neoplasms. Although the existence of false negative and false positive cases in the diagnosis by the prototype system suggested the necessity for a tuning of image composing procedures, the autofluorescence imaging video endoscopy is considered to contribute to non-invasive detection of neoplastic lesions of stomach.