Abstract: As the basis of food production, arable land is the root of the survival and development of human society, and its utilization efficiency has profound influence on the development and progress of human civilization. In this study, southwest region in china including Chongqing city, Sichuan province, Guizhou province and Yunnan province as a case study, and based on the dataset in 2000, 2005, 2010 and 2015, the DEA model was used to simulate the efficiency of arable land. Then, the global and local spatial pattern in different period of the arable land use efficiency had been analyzed by the method of global Moran’I, trend surface analysis and G*i index, respectively. At last, the main factors that influence the spatial differentiation of arable land utilization efficiency were evaluated by the method of geography weighted regression (GWR). The results showed that, firstly, between 2000 and 2015, the number of cities in high value areas of arable land utilization efficiency was expanding. Specifically, in the spatial distribution, the efficiency was stable, and the high value region mainly distributed in Dazhou-Deyang-Chengdu-Ganzi in Sichuan province. Sichuan province had maintained high efficiency and changed little; while, the efficiency in Chongqing city increased over time. The efficiency in Guizhou tended to decline, while Yunnan province remains inefficient. Secondly, the efficiency was mainly the positive spatial autocorrelation, and the spatial distribution trend increased from west to east, as well as the increase trend of the “U” pattern from south to north. Remarkable comprehensive of efficiency spatial pattern of differentiation, and relatively higher value/lower value of efficiency presented strong aggregation depended on the space distribution, cold/hot spatial pattern polarization phenomenon more obvious, clearly line and hierarchy. Change over time, the region of cold/ hot spots on the spatial pattern had changed from the “group” pattern to the “tape” pattern. Thirdly, the spatiotemporal change of the arable land utilization efficiency was the largest affected by driving factors of the per capita net income of farmers, followed by the multiple crop index; the effect of cultivated land quality and irrigation index was similar; the influence of per capita GDP to cultivated land use efficiency increased year by year; while, the effect of terrain factors decreased year by year.