This study was carried out in order to estimate site indices and to investigate productivity of Pinus koraiensis in unstocked land by the relationship between site index and physico-chemical properties of forest soil in stocked land. In Pinus koraiensis stand 83 nationwide sites were selected and site index model were estimated with 14 items of physico-chemical properties respectively obtained from A and B horizon of forest soil in each site. According to the relationship between soil properties and site index, it was found that the regressed models by soil properties fitted well in both site production prediction of soil A and B horizons. As a result of these two regressions, significant correlation was found between site index and Exch. Ca++, O.M., Clay, T.N., and Sand in A horizon, while Avail. P2O5, Base Sat., C.E.C., Exch. Mg++, and T.N. and site index have grater correlation in B horizon. In conclusion, these results will provide not only the important criteria for establishment of management forest plan in unstocked land, but also the useful guidance for selecting the suitable sites and trees.