The object of this paper is to clarify the factors of the spatial dispersion of affor-estation by simulation analysis. Two small areas which belong to two hamlets, Honmodani and Sakamotogawa, of Yusuhara-mura in southwestern part of the Shikoku mountainous region, were chosen for this study. The inhabitants of Honmodani are generally composed of smaller landowners and those of Sakamotogawa larger ones. Most of these two areas had been utilized for shifting cultivation and later for the cultivation of “mitsumata” (Edgewortia papyrfera) which was traditionally used for the production of Japanese paper until 1955. After World War II, afforestation has been extended in these areas, the change of which is shown in Figs. 3 and 11. In the application of simulation, a grid composed of 250 meter square meshes was overlaid and a probability field was prepared in each mesh as shown in Figs. 6 and 8. This probability field is based on the correlation between the relative number of the places of afforestation and the distance from the place of settlement or the places of afforestation distributed in the previous periods. So, Monte Carlo simulation is to be operated on this probability field. The results obtained are as follows: 1. In the area of I-Ionmodani, the trend of aforestation can be divided into four periods based on Fig. 3. In the 1st period, the distribution of actual output of affor-esta.tion is shown in Fig. 4-A and its simulation in Fig. 4-B. Spatial dispersion in this period is mainly related to the distance from the settlement. In the 2nd period, the distribution of actual output is shown in Fig. 5-A and its simulation in Fig. 5-B. From it, it is admitted that the afforestation in this period is related to the distance from the places of afforestation located in the previous period like the Markov chain. In the 3rd period, in which a greater part of the area was covered by afforestation, the actual and its simulated outputs are shown in Fig. 7-A, B. Spatial dispersion in this period is related to the distance from the places of afforestation located in the 1 st and 2nd periods like the cases of higher-order Markov chain. The actual and simu-lated outputs in the last period are shown in Fig. 9-A, B. In this period, it is also admitted that the spatial dispersion is based on the distance from the places of affor-estation distributed in the previous period. Through each period, a little differences between the distribution of actual output and simulated output are admitted. This is due to the distribution of the forest lands of the larger land possessors in this area. 2. In the area of Sakamotogawa, the trend of afforestation can be divided into three periods based on Fig. 11. The distribution of actual and its simulated outputs, in the 1 st and 2nd periods, are shown in Fig. 12-A, B and Fig. 13-A, B. Through these periods, it is admitted that the distribution of the places of afforestation is related to the distance from the settlement in the 1 st period, and the places of afforestation formed in the previous period in the 2nd period. But, in the 2nd period, contrary to the other periods, the increasing correlation between the ratio of the number of the places of afforestation and its distance from the places of afforestation formed in the previous period is admitted. This is due to the stability of the economic base of the farmers with larger forest lands and cultivated fields in this area. In the last period, the places of aforestation cover the area all over. 3. Thus, it is understood that the most important factor is the distance from the settlement in the 1st stage and from the places of the aforestation in the later stages, and the secondary factor is the distribution of forestland possessed by each landowner.