Detecting the determinants of spatio-temporal distribution of species is prerequisite for ecological conservation and restoration. Maximum entropy (Maxent) modeling was applied to investigate the present and future potential distributions of an endangered canopy-tree, Shorea robusta C. F. Gaertn. (Dipterocarpaceae), under urbanization in Purbachal, Bangladesh. The model was constructed by 165 location records that cover the whole distribution range of S. robusta.  Eight environmental variables in relation to climate, geography, and soil were included in the models.  Two scenarios proposed by IPCC (representative concentration pathways, RCP) were used for the prediction of distribution altered by global warming (from 2046 to 2065 and from 2081 to 2100). The accuracy of predicted distributions was supported sufficiently by the binomial test of omission (P ≈ 0.00) and area under the curve analysis (AUC > 0.97). The distributions were mostly determined by precipitation and soil nitrogen because S. robusta requires high precipitation and soil nitrogen because of canopy tree.  Maxent predicts that the suitable areas for S. robusta forest decline to 86.5% by 2100 in the climate change scenario of RCP8.5. Maxent results suggested that not only precipitation but also soil nitrogen should be monitored for the conservation of S. robusta forest.