The basis for monitoring the cultivated land quality is the sampling method. In order to improve monitoring accuracy of the cultivated land quality, the study built a spatial sampling method of cultivated land quality based on the monitoring control region spatial variability. This method calculated the optimal sampling number of the whole study area by Neyman method under a certain confidence and sampling error firstly. Then the whole study area was divided into several monitoring control regions based on the cultivated land quality and factors which affected the cultivated land quality. These monitoring control regions were used to stratified sampling. Samples distributing for every monitoring control regions based on their spatial variance coefficient were the next step. The spatial random sampling method was used to array these distributed samples in every monitoring control regions. 4 other kinds of sampling methods were used in this paper to contrast the estimation accuracy of this method include the spatial random sampling method, the stratified sampling method based on the physical quality grades of cultivated land and the stratified sampling method based on weights of monitoring control regions, etc. By contrasting the Kriging estimation error of these 5 kinds of methods, the estimation accuracy of different methods were different signally with the same number of samples, and the estimation error(measured by RMSE) of the spatial sampling method of cultivated land quality based on the monitoring control region spatial variability was smallest. In conclusion, a scientific sampling method can obtain the largest information and increase the efficiency and accuracy of the cultivated land quality monitoring.