The author considers that there should be a purpose for each regionalization of agricul-ture. From this viewpoint, he attempted an agricultural regionalization of Japan employing indices of farm-income changes as its data. The intended regionalization was achieved through three-step regionalization of 305 districts of Japan.<br> The 305 districts were classified on the basis of the rate of increase of farm income from 1960 to 1974 into three categories, high growth districts, medium growth districts and low growth districts (see the key for Fig. 2). The first-step regionalization, which produced 11 regions (Fig. 2), was based on the above classification of the districts and the spatial contiguity of the districts. For the second-step regionalization, an index of change in the crop combination of a district in terms of incomes from the crops was developed. On the basis of this index and the spatial contiguity of the districts, the 11 regions were divided into 23 subregions (Fig. 3).<br> For 1960 and 1974, the 305 districts were classified into four crop combination types: grains and potatoes, primarily grains and potatoes with commercial crops, primarily commercial crops with grains and potatoes, and commercial crops. Each district's shift from one crop combination type to another from 1960 (Fig. 4) to 1974 (Fig. 5) was categorized (Fig. 6). In the third-step regionalization, which was based on the classification of the districts shown in Fig. 6, the 23 subregions were further divided into 33 areas (Fig. 7). The result is an agricultural regionalization from the viewpoint of agricultural changes in Japan, which experienced rapid growth of its economy during the period. In the above regionalization, the agricultural changes were measured by the farm-income growth and the crop combination changes. The author assumed that crop combination reflected qualitative as well as quantitative aspects of farm management.<br> The unit areas (the 305 districts) used by other researchers were uncritically adopted in this paper, which might have been ill-suited for the purpose of this study. In drawing the regional boundaries, the problem of handling a unhomogeneous district or a few unhomo-geneous districts within a region was a difficult one and remains so. Further investigation of agricultural regionalization based on indices of changes is necessary. Moreover, since the increase of non-farm incomes of farm households is particularly rapid in Japan, agricultural regionalization based on indices of income-changes of farm households, which include non-farm incomes as well as farm incomes, should be studied. It is also desirable to analyze agriculture of small areas in order to understand the process of transformation of agricultural regions.