I) Gastric cytodiagnosis now plays an important role in the field of detection of early gastric cancer, in which infiltrations are localized in the mucosa and submucosa of the gastric wall. However no satisfactory results have been reported by several authors. As for the cause of this, two following problems should be discussed. 1) Difficulties in the method of collecting cells in early cancer of the stomach. 2) Cytological difficulties in diagnosis of exfoliated cells of early gastric cancer comparing to advanced ones. In this paper, we studied chiefly on the first problem. II) During 1961 and 1962 we detected 73 cases of gastric cancer, and in 49 cases we obtained. cytologically positive results (68%) by the ordinary lavage method. Out of them, only 4 case of early cancer and 2 of advanced carcinoma with small naked surface areas of diameter under 3 cm were detected. In 4 of the early cancers one was cytologically positive, one was in class 3, and the remaining 2 were negative. One of the advanced carcinomas with small naked surface area of diameter under 3cm was cytol ogically positive and another was negative. Since 1963 we have changed the method of exfoliative cytology and used a new method. After inserting a new gastric tube of hardened rubber (65 degrees of J. I. S.), with a lead-rubber at its tip in order to be seen fluoroscopically, suspected gastric leasions are locally lavaged repeated ly with proteolytic enzyme solution containing water-soluble contrast media (Urokolin). We call it “selective gastric lavge”. This method enabled us to lavage directly selected parts of the stomach under fluoroscopy. From Jan. 1963 to Apr. 1964, 57 cases of gastric cancer were examined with our new method. Among them 7 cases were of the early stage, and 4 were cytologycally positive, 2 in class 3, and 1 negative. There were 2 cases with small naked areas of diameters under 3 cm in the advanced cancers. One was positive and another negative. In other advanced gastric cancers, all were cytologycally positive. In every early cancers and small naked area cancers step-sections were made and histologycally determined. III) Until now the results of exfoliative cytology of gastric cancer have been said by many authors to be influenced by the location of the cancer. But according to our results, when the cancer has a large naked surface area, consistent positive diagnosis was obtained. On the contrary the smaller the naked surface area is and the lesser the cytologycal atypism is, the less positive results obtained. There may be several factors which make gastric cytology difficult: 1) Locations and forms of the stomach and compositions of the gastric contents (as blood, pH.etc.) 2) Inter-combinin strength of cancer cells. 3) Surface covering materials or membrane of cancerous ulcer, particularly by early cancer. 4) Largeness of the naked surface area of the cancer. 5) Cancer cell density of the surface area of the cancer. Of these we considered that the most important factors are largeness and cancer cell density of the neked surface areas. Other factors are almost excluded by performing our selective gastric lavage under fluoroscopy. We conclude that when the gastric lesions are suspected to be cancer of small naked surface (under 3cm in diameter) by fluoroscopic and endoscopic examinations, gastric cytology will be diffi cult to get positive results, so “selective gastric lavage” has to be done repeatedly. Sometimes endoscopic biopsy may be usefull.