Abstrac-Location of emergency logistics was strategic issues of Emergency Logistics System construction, the defect of traditional Emergency Logistics facilities model is a single evaluation,which use AHP to evaluate qualitative factors that affect logistics location and regard.AHP and DEA evaluation results as an input. DEA evaluation results act as a final policy recommendation.This method can increase the reliability of location and reduce the complexity of decision-making.China was one of countries that suffer the most serious natural disasters, diverse disasters, wide geographic distribution, high frequency, heavy losses characteristics of China national disasters. In these backgrounds, the location of Emergency Logistics location stands out rapidly. When selected emergency facilities will be long operating, efficiently and service levels have significant impact. Therefore, establishing reasonable Emergency location is strategic issues of the Emergency logistics system construction. Emergency Logistic [ ] 1 are provided emergency supplies for natural disasters, public health incidents, major accidents and other emergencies, in order to maximize time efficiency and minimize disaster losses. Emergency logistics and general logistics, have the spatial effects, time effects and shape mass effects. consisting of the fluid ,carrier, current direction, discharge, process, velocity of flow and other elements of composition. Location problem is one of most important issue in the logistics management system, although location have relatively mature model, emergency logistics site selection and the traditional location with different optimization constraints and goals, The former stressed the timeliness, the later is more concerned about costs. Therefore, before making the emergency logistics system, we must fully consider the reasonable sides of each emergency logistics, to enhance system efficiency, lower construction costs and lower operating costs. as for emergency facility location problems, domestic and foreign scholars have more researches. Odesd Berman and Dmitry Krass proposed maximum coverage model, which transfer in the largest traditional model of coverage dualistic assumptions diverse into assumptions, including "partial coverage" between completely covered and incompletely covered, through improved coverage, had better results than achieved maximum coverage. But this paper for coverage of quantitative criteria is still not clear, researches on the location of emergency facilities are not much. JohnHodgson pointed out that solving the P-median model requires two datas:1):the distance matrix;2):weights and facility quantity. And we get the actual data often have deviation with reality. He put several groups with different types of dates error rate respectively into the model, As a result the P-median model is Robustness, and the second class data error have more tolerance. Missing is that two sets of data don't input at the same time, nor consider the emergency limit period constraints of the P-robustness m...