The purpose of this study is to examine empirically the factors that determinate forest household to participate in reforestation program by forestry censuses carried out in 2000 and 2005 respectively. An analysis model is developed mainly based on land rent theory to find out the factors. The household data of censuses are regrouped into township level data in order to run regression analysis. A spatial analysis is applied to detect and modify the spatial effect of different townships. The study result indicates that the age of householder, forest income generated per self-own labor, the proportions of second and third industries in township and the participating rate in 2000 discourage the participation in reforestation program in 2005. However, the level of education, proportion of flat land, distance of forestland to local market, proportion of forestland owned privately and the amount of subsidies have the positive tendencies towards participating reforestation program .