BACKGROUND/AIMS: Magnifying colonoscopy was developed for detailed examination of the surface of colorectal neoplastic lesions. While magnifying colonoscopy is useful for differentiating neoplastic lesions from nonneoplastic ones, for evaluating early colorectal cancers, it still has limits in practice. This study was designed to clarify the usefulness and the limits of magnifying colonoscopy. METHODS: Three hundred and fifty-two lesions, which were observed using magnifying colonoscopy from July to August 1999 and whose histologies were proven, were analyzed according to their pit patterns. The pit patterns are I, normal round pits; II, large starry-shaped pits; IIIs, small round pits; IIIL, large round or rod-shaped pits; IV, branched or gyrus-like pits; and V, irregularly shaped or nonstructural pits. In cases where a pit pattern was hard to read, the pattern was classified as 'D'. RESULTS: The dominant pit pattern for protruded lesions was IIIL, accounting for 44.6%. In diminutive lesions (< or =5 mm), II and IIIL were equally common, 40% of the total for those lesions, respectively. In medium-sized lesions (from 6 to 10 mm), IIIL was the major pit pattern, 45.6% of the total for that type of lesion. In lesions larger than 10 mm, IIIL and IV were the most common pit patterns, each accounting for 26.7% of the total. The overall accuracy ratio of pit pattern diagnoses was 79.5%. The frequency of difficult pit patterns to read was 6.3%. Among them, 77.3% were due to difficulty in interpreting the pit patterns, and 22.7% were due to an inability to clarify the pit pattern because of poor staining. Ninety-one percent of the difficult cases to read involved diminutive lesions, and 86% of those difficult cases involved tubular adenomas. CONCLUSIONS: The 80% accuracy rate for pit pattern diagnosis suggests that magnifying colonoscopy is probably useful, but problems, such as poor staining due to mucus and difficulty with interpretation, still exist.