To understand the spatial pattern of cultivated land productivity in a certain region, a model for evaluating such productivity was established by analyzing the relationship between the cultivated land quality ranks and the standard crop yield per hector through sampling investigation data on the basis of agricultural land classification. With this model, spatial patterns of theoretical, accessible and actual productivities in Tangshan were assessed. The region with higher theoretical and accessible productivity was centered from Tangshan city, while the region with higher actual productivity was located at the south and northeast area of Tangshan. The hilly areas including Qian’xi and Qian’an were characterized with relatively low productivity for all three levels. The west, east and southeast areas in Tangshan were associated with relatively high total productivity but hilly area in north Tangshan with relatively low total productivity. The results based on the exploratory spatial data analysis (ESDA) spatial autocorrelation indicated: the accumulation order of three levels productivity was theoretical productivity per unit > accessible productivity per unit > current productivity per unit; in addition, the total area in H-H (high spatial autocorrelations) and L-L (low spatial autocorrelations) type accounted for over 50% of Tangshan city. It also indicated that there are good spatial correlations among accessible productivity, GDP per capita and net income of rural people as well as good spatial correlations between actual productivity and farming production. This is very useful for the sustainable use of cultivated land and development of local economy by understanding spatial pattern of regional cultivated land productivity and its relationships with the development of regional economy.