Abstract: In this paper, a comprehensive indicator system of agricultural green total factor productivity (TFP) is proposed based on the concept of total factor productivity, and the corresponding agricultural green TFP for the Yellow River Basin provinces from 2004-2017 is calculated using the SBM model. The spatial autocorrelation analysis, kernel density estimation and Markov chain probability matrix are used to analyze the spatial pattern and dynamic evolution of the proposed indicator system. The results show that the agricultural green TFP in the Yellow River basin is not high as a whole, the trend of change between 2004-2017 was slightly decreased. There are obvious differences between provinces, and the spatial connection of the agricultural green TFP in the Yellow River basin is weak. The high-level and low-level agricultural green TFP areas are agglomerated together, which forms "club convergence". Under the effect of spatial lag, the phenomenon of "efficiency locking" was displayed. This study provides a policy implication for promoting the coordinated development of the Yellow River Basin and the green development of agriculture.