To clarify the actual conditions and problems encountered in the diagnosis of gastric cancer during routine endoscopic examinations using electronic endoscopes, a retrospective study of cases that had shown inconsistencies between the endoscopic and biopsy diagnoses was conducted. The subjects were 5,640 patients who had undergone gastric biopsies. They were selected from among 18,248 patients in whom upper gastrointestinal endoscopy had been performed between July 1992 and July 1997 at the National Cancer Center Hospital East. The study was conducted focusing on macroscopic type, size, location, and pathologic findings of the lesions. Cases showing inconsistencies between the endoscopic and biopsy diagnoses comprised 2.7% of all biopsy cases (150 of 5,640). Most of these inconsistently diagnosed lesions were less than 20 mm in diameter. The frequency of inaccurate diagnosis for depressed lesions in the areas of upper and middle thirds of the stomach from the lesser curvature to the posterior wall was significantly higher in the inconsistent group, compared with the accurate diagnoses from the control group. In diagnostic reviews of depressed lesions, 46.0% (46 of 100) were diagnosed accurately, but cases with fewer malignant indices were recognized. These results indicate that diagnostic accuracy must be enhanced through objective biopsy-implementation standards and the use of image processing in using diagnostic methods that employ electronic endoscopes.