Abstract: To model the potential distribution of Bretschneidera sinensis Hemsl. and compare the spatiotemporal population dynamics of B. sinensis under Last Glacial Maximum, Mid Holocene, Current, and Future periods (RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5), 151 existing populations of this species and 12 climate variables were selected based on the MaxEnt model and GIS. Using the receiver-operating characteristic curve (ROC), we analyzed the credibility of the MaxEnt model. To identify the climate variables restricting the distribution of this species, we used training gain, percentage contribution, and permutation importance of the climate variables. The spatiotemporal population dynamics of B. sinensis were compared using the distribution area ratio (N_a) and extent of habitat change (N_e), respectively. The area under the ROC showed that the AUC values of the training data and test data under the seven different climate scenarios were higher than 0.99. This indicated that the simulation accuracy of the MaxEnt model was extremely high. Training gain, climate variable contribution, and permutation importance revealed that the geographical distribution of B. sinensis was limited by mean diurnal range (mean of monthly), isothermality, and precipitation of the driest quarter. The spatiotemporal population dynamics of B. sinensis from the Last Glacial Maximum to Mid Holocene indicated that the Jinfoshan and Dayaoshan mountains likely provided micro-glacial refugia for this species, after which geographical distribution experienced an expansion. In the future, the potential geographical distribution of B. sinensis may diminish from 25%-46%, depending on changes in greenhouse gas emissions, with the range of B. sinensis being the smallest and most fragmented under RCP 8.5. Implementing international cooperation in the investigation, collection, and genetic management of this species, and establishing ex-situ forest gene conservation in climate friendly areas could be important ways to protect and restore this species.