Flavor is an essential quality characteristics of soymilk, which contains volatile compounds derived from
fatty acids via enzymatic and thermal reactions. In this study, 67 kinds of soybean cultivars were selected,
and correlation analysis was conducted between physicochemical indexes of these soybean cultivars and
flavor characteristic indexes of soymilk. With clustering analysis, all the soybean cultivars could be
classified into three classes, and according to the results of principal component analysis for each class
of soymilk flavor characteristics, the soymilk of second class had relatively heavier beany and non-beany
flavor, and the third class had weaker flavor. For soybean cultivars of which the soymilk characteristics
were unknown, two discriminant functions could be used to predict flavor characteristics if the physicochemical
indexes were known. Therefore, screening of soybean cultivars suitable for soymilk processing
can be targeted for the flavor favored by consumers and an evaluation method established.