Based on the fundamental data of R&D input and output in China's 30 provinces and autonomous regions (excluding Tibet) since 2009, the super efficient DEA model is used to measure the efficiency of government R&D funding in various regions. The results show that there are significant inter-provincial differences in efficiency of Chinese government R&D funding for reasons of funding scale and management. Then, in order to further reveal the deep-seated reasons for this difference, the Tobit model is used to measure the efficiency of government R&D funding. Empirical analysis of the reasons at the enterprise level shows that the government R&D funding rate, the stability of the R&D funding policy, the government R&D funding allocation structure, the R&D personnel qualification structure and the R&D investment intensity all have significant effects on the government R&D funding efficiency. Finally, on this basis, the corresponding policy recommendations are proposed for the improvement of the efficiency of Chinese government R&D funding.