Land finance has been favored by local governments along with the rapid expansion of urbanization and industrialization, but it draws the attention of the central government recently and then tends to be alleviated. The influencing process of land finance on economic efficiency is systematic and complex. Considering the structural characteristics of economic efficiency growth, this paper constructs the multi-level mechanism and path of land finance's influence on economic efficiency, from the aspects of scaling effect, technical-inhibition effect, investment crowding-out effect and cost-pulling effect. Then this paper employs the fixed effect model to empirically test the impact and regional differences of land finance on total factor productivity (TFP) of 285 Chinese cities from 2005 to 2015. The results show that: Firstly, China's urban land financial scale continued to expand, and spread from the coastal cities to the inland cities in the study period. At the same time, the urban TFP growth rate presents a fluctuation downward tendency. Secondly, the baseline regression shows that the land finance expansion has a significant negative impact on the growth of urban TFP. And the impact on western, central, northeastern China and small and medium-sized cities, I type and II type cities gradually increase in order, while the impact is not significant in eastern China and large cities with a population of three million or more. Thirdly, although the land finance has a certain scale economic effect on regional development, it mainly inhibits technological progress to directly hinder the TFP growth. And it is more serious in central and western China and cities with a population more than one million. Further, it is mainly through investment crowding-out effect brought by the real estate development that the land finance affects urban TFP, followed by the cost-pulling effect by house prices rising, while wages rising mainly affects urban TFP in eastern China and cities with a population less than three million. The above conclusions are still valid after a series of robustness tests. Finally, this paper puts forward some policy suggestions, such as defining the scope of revenue and expenditure of land finance, gradient transfer and upgrading of industries and macro-control and implementing targeted policies to each city.