Taking 171 counties of Beijing-Tianjin-Hebei region as research units, adopting GIS spatial analysis methods, it is revealed that spatial difference of agricultural labor productivity in 1994, 2000, 2006 and 2012. With geographically weighted regression model, the causes for the spatial difference of labor productivity in 2000 and 2012 are revealed. The results indicate that the agricultural labor productivity at county level shows unbalanced development with remarkable special differentiation. The counties in Beijing-Tianjin-Tangshan region possess higher agricultural labor productivity, however, there is a slow increase in labor productivity for the counties in Beijing, obvious decrease in number of agglomeration unit. The agricultural labor productivity of the counties in Shijiazhuang surrounding area sees high-level agglomeration; Agricultural labor productivity of the counties in Zhangjiakou, Chengde, Baoding and Xingtai is situated at a relatively low level. During the research period, agricultural labor productivity has a rapid increase, with no obvious polarization trend. In four research years, agricultural labor productivity at county level shows positive correlation but with weakened agglomerating level, so agricultural labor productivity at county level shows a decentralized sign. Simulation result of geographically weighted regression model is significantly better than ordinary least squares. Parameter estimation results for regression coefficients of controlled variables of 171 countries are different. Driving factors of labor productivity of agricultural work are featured as localization other than unbalanced linkage, and effects of agricultural labor productivity in previous stage are most obvious. Therefore, current status of agricultural labor productivity and driving factor should be combined to optimize agricultural labor productivity in Beijing-Tianjin-Hebei region.