To evaluate the role of Lugol dye endoscopy in diagnosing early esophageal cancer, we reviewed findings of dye endoscopy and those of conventional endoscopy in 17 early esophageal cancers that were demonstrated as unstained areas on dyeing with Lugol solution. Histologically, all 17 lesions were squamous cell carcinomas; 10 lesions being mucosal carcinomas, the remaining 7 lesions mucosal carcinomas spreading beyond the epithelial layer. The lesions ranged from 0.7 to 4.0 cm in size. Abnormal findings were noted under conventional endoscopy in all but 3 lesions diagnosed only by postoperative pathohistology, regardless of the size and depth of the invasion. Under conventional endoscopy, the following types of morphological changes were noted in 8 (57.1%) of the 14 lesions: slight elevation (1 lesion), depression (6 lesions), and deformed arc (1 lesion). A color change was noted endoscopically in 12 of the 14 lesions (85.7%), this change being redness in all 12 lesions. The unstained area on the resected specimen was consistent with the size of the lesion that was determined by using serially sectioned blocks in all cases. Moreover, the former completely (100%) coincided with the histological area where PAS reaction was weak. In conclusion, under conventional endoscopy, a color change such as redness is an important indicator of minute or superficial esophageal cancer, as is such morphological change as depression, elevation or deformed arc. On the other hand, Lugol dye endoscopy is very helpful in detecting esophageal cancer unassociated with any morphological or color change. It also provides accurate information about the extent of the cancer.