The aims of this paper are to identify inter-prefectural migration patterns, and to extract “migration fields” which are defined as joint origin-destination regions of migration (Schwind, 1975). Data for this study were taken from the Resident Registration data published by the Bureau of Statistics, Office of the Prime Minister, Japan. The data are expressed in the form of 46×46 origin-destination migration matrix. In the first stage of the study, principal component analyses were employed to reduce the above-mentioned migration matrix to its underlying dimensions, representing regional groups of places (prefectures) treated as destinations and origins. The results are as follows: Seven dimensions with eigenvalue of greater than 1.0, explaining about 83 percent of the total variance of the 46×46 migration matrix, were identified in the R-mode analysis, for the period of 1960-65 (Fig. 1). Dimensions in the R-mode analysis represent regions composed of prefectures (indicated by component loadings of greater than 0.5) having similar patterns of inmigration from other prefectures. Figures 2-a_??_c show the results of these dimensions. On the other hand, six dimensions with eigenvalue of greater than 1.0, explaining about 91 percent of the total variance of the migration matrix, were identified in the Q-mode analysis, for the period of 1960-65 (Fig. 1). Dimensions in the Q-mode analysis reveal regions composed of prefectures having similar patterns of outmigration to other prefectures. Figures 3-a_??_c show the results of these dimensions. In the second stage, canonical analysis was employed to identify the interdependencies of the inmigration (destination) regions and outmigration (origin) regions, and then to test whether migration fields defined above can be identified within Japan, for the period of 1960-65. The results of the canonical analysis are summarized in Table 2. All canonical correlations were found highly significant by the chi-square test. And five pairs of canonical variates were formed from the migration dimensions of seven destination regions and six origin regions. So, each pair of canonical variates represents a migration field of prefectures with similar inmigration and outmigration patterns. By the empirical interpretation of each migration field, five migration fields were extracted as Eastern Japan field, western Japan field, Yamaguchi-Kyûshû field, Chûkyô field, and Ôsaka field. How, then, did the inter-prefectural migration patterns change, for the period of 1966-70? So, the above-mentioned same analyses were applied to the data on inter-prefectural migration stream, for the period of 1966-70. The part of results of analyses are illustrated in Figures 4_??_6 and Table 3. For the period of 1966-70, five migration fields were extracted as Eastern Japan field, Western Japan field, Yamaguchi-Kyûshû field, Ôsaka field, and Tokyô field.