We described the associated variation in site index of Pinus densiflora for. erecta in responses to site variables of soil properties. Additionally, this study was conducted to estimate site index in unstocked land by the physico-chemical properties derived from stocked forest soil and to investigate the effects of the physico-chemical properties by soil layers on site index of surveyed area, totally 77 sites in central Korea. At each site, there were surveyed 14 items of the physico-chemical properties, namely, content of sand, silt and clay, soil pH, organic matter, total nitrogen, available P2O5, cation exchange capacity, exchangeable K+, Na+, Ca++, Mg++, and base saturation. The multiple regression equations using stepwise regression method were applied to this study. In the degree of contribution of the variables to site index by soil horizons, the highly effective variables were Base Sat., Exch. Ca++, T.N., Exch. Mg++, and C.E.C. in A horizon and Base Sat., Exch. Mg++, C.E.C., Soil pH, and O.M. in B horizon of P. densiflora for. erecta stands. As a result, it was concluded that the regressive prediction induced by significant soil site factors by forest soil horizons had useful guidance for finding out the best suitable area especially in unstocked land and further more for establishing successful stands of this species.