Spatial stratification is a key link in the monitoring and evaluation of cropland quality. Taking Horqin Left-Wing Middle Banner as an example, a method for spatial heterogeneity stratification in arable cropland quality was proposed based on categorical principal component analysis. The weights of cropland quality indicators were quantified by categorical principal component analysis method, and the indicators were classified into important and general indicators by combining the mean-standard deviation grading method. The important indicators were classified by using a classification and gradation method according to the corresponding national standard, and the spatial clustering stratification of the general indicators were carried out based on the two-step clustering method. The spatial heterogeneity stratification of cropland quality was achieved by constructing the stratification and fusion rules. The stratification results were evaluated in both quantitative and qualitative terms by using geographical detector and consistency testing method, respectively, and were compared with the full-indicator clustering stratification method. The results showed that the indicators of cropland quality in Horqin Left-Wing Middle Banner were divided into important indicators and general indicators, and the results of spatial heterogeneity stratification for cropland quality were high, medium-high, medium, medium-low, and low strata, and the corresponding proportions of stratified areas were 12%, 22%, 28%, 24% and 14%, respectively. The q values of the stratification results calculated using the geographic detector were 0.74 and 0.67, and the stratification results were consistent with the evaluation data of cropland quality and the grades of cropland quality. The effect of spatial heterogeneity stratification was better than that of the full-indicator clustering stratification method. The developed spatial heterogeneity stratification method of cropland quality can highlight the contributions of important indicators to cropland quality, and provide a technical support for rapid screening and dynamic monitoring of cropland quality. The developed method for spatial heterogeneity stratification in arable cropland quality further highlighted the contribution of key indicators to soil fertility, thereby providing technical support for rapid screening and dynamic monitoring of arable cropland quality.