This study focused on Taiwania plantations on forest land of the Danaen working circle in the Donshu National Forest. The normalized difference vegetation index as greenness of each stand was calculated from a SPOT image. Four forest habitat factors, temperature index, synthetic moisture gradient, whole sky space, and direct light incidence ratio, associated with topography and location were derived from digital elevation data. In addition, the Taiwania stand age, crown closure, and volume were selected from the forest land-use geographic information data base. Greenness, 4 forest habitat factors, stand age, and crown closure of the entire area of each stand were independent variables, and stand volume as was the dependent variable in the multivariable stepwise regression analysis which searched for the best volume estimate model. Through stratifying the terrain configuration factor derived by the digital terrain model, relieving the higher relief stratification from each stand stepwise, R^2 of the best model was able to reach 0.80. This shows that the Taiwania volume estimate models are affected by the variance of terrain. According to how easy it is to acquire the dependent variables, the optimal timber volume estimate models can be selected by the analysis procedure in this study.