Napier grass is planned to be planted as an energy crop in Thailand by the Ministry of Energy who launched the Alternative Energy Development Plan (AEDP 2012-2021) to increase the use of alternative energy. To effectively plant Napier grass, the influence of geographic factors, which are soil texture, soil drainage, soil fertility, soil pH, and precipitation, on the Napier Grass yield has to be quantitatively understood. Although the multiple linear regression analysis for predicting the Napier grass yield has been carried out, it does not take the non-linearity of the data as well as the interaction between the factors into account. In this study, the multiple regression analyses for Napier grass dry matter yield prediction, with and without considering the effect of nonlinearity and interaction between variables, were studied and compared to investigate the accuracy between the analyses. The results demonstrated that based on the 105 sets of data collected from different pilot plants throughout Thailand, the multiple polynomial regression analysis with the interaction variables provided an accuracy of 81.0% at the confident level of 95% which was higher than that of the linear regression which has no effect of non-linearity.