In Japan, the movement from rural prefectures to metropolitan prefectures has been a major migration stream since the middle of Meiji Era when the industrial revolution started. This long-standing trend, however, has changed recently. The prominent stream is the migration from metropolitan prefectures to rural prefectures which is reverse of the long-standing trend.<br> This stream is called “U-turn” which means return migration. We have a lot of discussions about it. This subject may be investigated in the broader context. The author aims at detecting the epoch of internal migration in postwar Japan and the major factors which affect the recent internal migration including “U-turn” phenomenon.<br> This study, first, examined the change of the net migration rates for prefectures shown in “Annual Report on the Internal Migration in Japan Derived from the Basic Resident Registers” (Table 1). By the application of T-test method which discerns statistically the significant difference between two mean values, it is detected between the former and the latter halves of the 1960's. Namely, it was not until the latter half of the 1960's that the internal migration in Japan has shifted to a new stage. Out-migration from rural prefectures became weaker, so that population concentration towards metropolitan prefectures was less distinct than before.<br> Secondly, major factors that affect the inter-prefectural migration were examined by multiple regression analysis. The years of 1955 and 1970 were chosen for the analysis, before and after the epoch, and all the 46 prefectures were grouped into two categories, i.e. metropolitan (10 prefectures) and non-metropolitan (36 prefectures). The former consisted of Keihin (Saitama, Chiba, Tokyo and Kanagawa), Chukyo (Gifu, Aichi and Mie) and Keihanshin (Kyoto, Osaka and Hyogo).<br> The inter-prefectural migration movements were summed up in the following four types;<br> type A. inter-prefectural migration in total<br> B. migration from non-metropolitan prefectures to metropolitan prefectures<br> C. migration from metropolitan prefectures to non-metropolitan prefectures<br> D. inter-metropolitan migration<br> From Tables 3 and 4, following findings have been obtained.<br> 1) For 1970, the coefficients of determination for all four migration types show decline. Therefore, recent migration is affected by more complex and numerous factors.<br> 2) Inter-metropolitan migration (Type D) is affected by more limited factors than types B and C, namely migration between metropolitan and non-metropolitan areas.<br> 3) The factor of inter-prefectural difference of per capita income is less valid for explaining all of the four types mentioned above, however, employment increase in destination still plays an important role. So-called “decline of economic motivation” might be partly approved.<br> 4) Distance is a strong and stable factor. Figures 1_??_4 show that migration is considerably determined by contiguity.<br> 5) The factor of accumulation of the higher central functions in major cities of prefectures in destination, which is considered as a leading factor for city growth, does't always play a significant role in all of the four types of migration.<br> 6) Other important factors are the psychological attitude towards migration behavior indicated by both a sum of past migrants for destination and a propensity not to migrate. The migrant's attributes represented by age and education are also important.<br> Furthermore, multicollinearity, that is one of several troublesome problems in application of multiple regression analysis, is examined by Riddell's method. However, it is clear that this problem does not distort original regression model used in the analysis as seen from the Table 4.