The cytological diagnosis of gastric cancer has been made rapid progress with the improvement of methods of cell collection. The knowledge of morphological characteristics of exfoliative cells in gastric cancer has been analysed in more details, however, the classification of cytodiagnosis depends upon routine morphological findings, which is different in different individual experiences of cytology. The is the reason why the objectivity of classification is required. The present study was aimed to the statistical approach for the purpose of establishment of standardization in cytodiagnosis of gastric cancer. Materials were obtained from resected stomach of 140 cases of gastric cancer, 140 cases of peptic ulcer, 20 cases of gastric polyp and 30 cases of gastritis. Three smear specimens were made in each case and stained by Papanicolaou's method. In the statistical analysis of morphological findings of exfoliative celles of these cases, the results as follows were obtained. 1 As the criteria of malignancy, following twelve findings were usefull and highly significant; crowding of cells and nuclei, irregularity of pattern, anisokaryosis, pleomorphism of nuclei, large cytoplasmic vacuolization, enlargement of nuclei, increase of nucleo-cytoplasmic ratio, irregularity in nuclear outline, thickening of nuclear membrane, chromatin condensation, enlargement of nucleoli, increase of nucleoli. 2 There were no statistically significant cytological findings between early cancer and advanced cancer. 3 Statistically, the difference of mean numbers of cytological findings between cancer group and benign group was significant. 4 When trial statistic diagnosis by the use of above mentioned twelve findings was performed, the frequency of incorrect diagnosis were 2.1% in likelihood method, 3.3% in discriminant function method and 2.7% in simple mumbering method of cytological findings, respectively. 5 The most relevant combination of cytological findings was nuclear pleomorphism and increase of nucleo-cytoplasmic ratio. 6 The source of incorrect diagnosis or borderline cases were mainly narrow exposed carcinoma simplex or early cancer. In benign group, on the other side, the main source of incorrect diagnosis was peptic ulcer with high grade metaplastic gastritis.