Revealing regional distribution and diversity of abundant and rare bacterial communities in different cropping systems are crucial to predict trends of microbial changes in farmlands and understand ecosystem functions. Here, we examined the spatial distribution patterns and calculated differences in diversity and function for abundant and rare bacterial sub-communities across 114 typical dryland farmland (corn, soybean and alfalfa) soils in crop monoculture (CM) and rotation (CR) systems in northeastern China. Significant (P < 0.001) distance-decay relationships were detected for the abundant and rare sub-communities in both CM and CR systems. In addition, edaphic variables (CM_abundant = 27.3%, CM_rare = 9.1%; CR_abundant = 33.3%, CR_rare = 15.3%) largely contributed to the variation of sub-communities based on variance partitioning analysis (VPA). Shannon and Chao1 indices were significantly (P < 0.05) different between abundant and rare sub-communities in CM and CR systems. Both abundant (R_ANOSIM = 0.065, P = 0.004) and rare (R_ANOSIM = 0.054, P = 0.013) bacterial community were significantly (P < 0.05) separated according to CM and CR systems. Based on Tax4Fun, for Proteobacteria, environmental information processing was the main potential function in CM, while metabolism was the one in CR system. Additionally, the potential functions of rare taxa were greater than that of abundant taxa both in CM and CR systems. Our results would not only provide theoretical support for better understanding the roles of rare and abundant bacteria, but also offered new opportunity to improve agricultural ecosystem functions.