Protecting cultivated land is to protect the foundation of development and ensure the food safety in China. Permanent basic farmland is an effective method to prevent the high-quality cultivated land from occupying by construction land, but also the basis for the construction of high standard farmland. At present, it is urgent to implement the strictest protection for cultivated land resources in China. A method for the determination of permanent basic farmland was proposed, which was based on TOPSIS and local spatial autocorrelation. Firstly, four aspects of cultivated land were considered to structure the all-round comprehensive quality evaluation system. Secondly, TOPSIS was chosen to evaluate the comprehensive quality of cultivated land. On this basis, the spatial clustering was joined, and the comprehensive quality of cultivated land was analyzed. According to the characteristics of local spatial autocorrelation type, the zoning rules were formulated. Finally, according to the comprehensive quality of cultivated land and the spatial clustering characteristics, the satisfactory cultivated land were classified into permanent basic farmland, and suggestions for protecting permanent basic farmland were provided. The new method was illustrated by using Gaobeidian country, Hebei province as a case study. According to the results, TOPSIS and local spatial autocorrelation were used, combined with the comprehensive quality of cultivated land and their spatial clustering characteristics, the former three types of cultivated land were classified as permanent basic farmland, the area of the permanent basic farmland in Gaobeidian was 339.61km2, the area ratio was 81.75%, which achieved the relevant state regulations. Finally, three aspects of management suggestions to permanent basic farmland were proposed, which were “make laws, strict supervision”, “careful selection, expand investment” and “strengthen education, give encouragement”. The ecological landscape factors were added into the quality evaluation of cultivated land, and combined the spatial clustering method which was based on the optimal distance threshold and the improved distance spatial weight matrixes, and it provided a new way for the permanent basic farmland.