A study of factorial ecology is one of the most useful techniques for the analysis of urban residential structures. Studies of the residential structure of Japanese cities using this technique have recently become popular. However, these studies do not deal with historical change in the ecological structures of the various factors.With this problem in mind, the following study was conducted: As it has been the largest city in Japan, Tokyo was selected as the most appropriate study area. The Taisho era was selected as the first step for historical analysis. During this period, the urban structures of the Castle-Town, which had been built during the Edo era, coexisted with modern structures built during and after the Meiji era.It has been said that the urban structure of Tokyo is generally divided into two areas: The white-collar residential districts (Yamanote) of the uplands, and the blue-collar residential districts (Shitamachi) of the fluvial lowlands. The areal pattern of the boundary has not, however, been investigated in detail until now. The purpose of this paper is to explain the urban residential structures of Tokyo during the Taisho era.In the first stage of the analysis, a set of variables describing the social and economic characteristics of residential groups in various districts of the city was selected from some population characteristics taken from the 1920 Population Census of Japan. A total of 19 variables and 816 districts were used to determine the socioeconomic factors shown in Table 1. The principal factor method was then applied to the 816 x 19 original data matrix, and factor loadings, factor scores, and eigenvalues were obtained. These factors were rotated by the normal varimax method in order to obtain a simple structure. Next, factor loadings were interpreted and factor scores were mapped.The factor analysis produced six factors which together accounted for 74 percent of the total variance of the data (Table 2). The first factor, accounting for 26.5 percent of the variance, separates the households of middle-aged craftsmen with children from wholesale and retail workers' household. Districts which were characterized by the households of middle-aged craftsmen with children extend to the eastern parts of Tokyo. For example, many districts in the Asakusa, Honjo, and Fukagawa Wards, and the western small valley tracts are included here (Fig. 2-a). High negative factor scores showing large wholesale and retail workers' household coincide with commercial districts (especially: wholesale store districts) in the Nihonbashi and Kyobashi Wards along the Tokaido and Oshukaido thoroughfare (Fig. 2-b). Negative factor scores are also seen in the western upper residential quarters. The second factor is interpreted as indicating workers in civil service and indepedent business enterprises. The districts having factor scores greater than 0.5 lay in Kojimachi, western Kanda, and the area from Akasaka to Hongo Wards, where sector patterns coincided with the Yamanote upland (Fig. 3). The third factor is interpreted as the “old widow factor”. High positively scored districts were distributed in the western residential areas and central entertainment districts (Fig. 4). On the other hand, high negative scoring districts were composed of four types of quasi-households. These are central wholesale and retail districts, eastern manufacturing areas, students' quarters and army quarters. The fourth factor is interpreted as blue-collar, the fifth factor as female workers and the sixth factor as unmarried persons. In cluster analysis the similarity between districts was measured by Euclid distance, and the hierarchical groupings were selected by the centroid method.