Assessing the contribution of various factors to the spatial variation of crop yield is of vital importance in promoting agricultural practice according to local conditions, such as delimiting functional regions of grain production, optimizing cropping system, improving medium and low yield fields, and developing high-standard grain fields. This study was carried out in Henan Province, China, by using the field fertilization experiment data at the county level from 2008 to 2010, cultivated land productivity evaluation data, and meteorological data in summer maize growing season. Employing the Boruta algorithm and random forest model, contributions of five factor categories (including 17 factors) namely fertilization, soil, climate, cultivar, and irrigation and drainage, to the spatial variation of summer maize yield were quantitatively assessed. It was demonstrated that the high-yield areas of summer maize were mainly distributed in the Huang-Huai-Hai Plain (HHHP) in the north, and the medium-yield fields were concentrically located in the HHHP in the east and in Nanyang Basin, while the low-yield parcels were scattered in the mountainous and hilly areas in the west, the south, and the periphery of Nanyang Basin. The fields with less inter-annual fluctuation of yield were mainly found in medium-yield areas. All the 17 factors played an important role in the spatial variation of summer maize yield. Fertilization, particularly formulated fertilization, effectively increased crop yield, however, it was not work on changing the spatial distribution pattern of crop yield. This led to the conclusion that the spatial distribution pattern of crop yield is dominated by regional factors such as climate, soil and site conditions. Thus, for sustainable food security, it is necessary to protect high quality cultivated land resources with high and stable yield, and to improve cultivated land productivity by eliminating soil obstacle factors and improving the site.