Analysis of stimulus similarity judgment is suggested to be used for the research of individual peculiarities of psychical processes. Critical discussion of the existing models of multidimensional scaling of similarity is reported. It is concluded that distance and content models have serious shortcomings limiting their application. The feature model proposed by A. Tversky and the classificatory theory which makes no dimensional or metric assumption are discussed. The author adheres to a statement that judgments of similarity and difference are complementary.