257 stomach cancer patients and 766 controls were interviewed using a standardized questionnaire for 35 variables. The two groups were compared by multivariate variance analysis. This analysis results in a combination of 20 variables significantly connected with gastric cancer. The rankorder of the most prominent variables is as follows: no food intolerance, higher alcohol consumption, living place near Berlin, occupation labourer, early loss of teeth, car driving, no gallstones, smoking and age. It can be assumed that these variables are of aetiological importance. In the aetiology of gastric cancer, exogenous cancerogens seem to play a more important role than endogenous factors. In two further multivariate variance analysis, control group is divided into persons with and persons without chronic gastritis verified by blind gastric biopsy and compared with gastric cancer patients, too. A fourth analysis is limited to female only. The results of the three last mentioned analysis are similar to those of the first one and also similar to results from other countries based on univariate analysis. The computed combination of variables was used for diagnosis of gastric cancer by discriminance analysis in the persons forming the material for this study. This diagnosis was correct in 71% of gastric cancer patients and in 72% of control persons.