【Objective】Fertilization is an effective measure to improve soil fertility and increase crop yield in red soil. As important components of a soil ecosystem, soil microorganisms play an essential role in soil fertility formation, organic matter decomposition, and nutrient biogeochemical recycling as well. However, though soil microorganisms are very sensitive to fertilization, limited information is available in the literature about effects of fertilization on the microbial community in upland red soil. The objective of this study is to investigate responses of the bacterial community in upland red soil to long-term fertilization varying in strategy, and to provide a theoretical basis for rational fertilization and sustainable utilization of the red soil in South China.【Method】In this study, a long-term field fertilization experiment (established in 1990) on winter wheat–summer maize rotation at the Qiyang Red Soil Experimental Station of the Chinese Academy of Agricultural Sciences was carried out. The experiment had four treatments, i.e. no fertilizer (CK), only application of chemical nitrogen fertilizer (N), application of chemical nitrogen, phosphorus and potassium fertilizer (NPK), and application of NPK plus manure (NPKM). Except CK, all the treatments received N 300 kg•hm^-2•a^-1, and the N in Treatment NPKM had 30% coming from chemical fertilizer and the rest from manure. Samples of surface soil (0~20 cm) were collected in May 2015 (after wheat was harvested and before maize was planted). The technology of high-throughput sequencing of the V4-V5 16S rRNA gene region were used to determine composition and diversity of the soil bacterial community in the samples, and then to identify key factors affecting the bacterial community in the upland red soil.【Result】Results show: (1) Long-term fertilization significantly changed chemical properties of the upland red soil. Compared with CK, Treatments N and NPK declined significantly or down to 4.02 and 4.15, respectively, in soil pH, while Treatment NPKM increased significantly or up to 5. 99. Both NPK and NPKM significantly improved soil fertility, but the effect of the latter was much better than the former. However, Treatment N only increased soil TN content, which was not conducive to soil fertility; (2) Long-term fertilization affected relative abundance of the soil dominant bacteria at the phylum level. Non-metric multidimensional scaling (NMDS) and similarity analysis (ANOSIM) shows that soil bacterial communities in the treatments varied significantly; (3) Compared with CK, Treatment N was 21.4%~49.4% lower in the four diversity indices (observed species, Chao1 index, phylogenetic diversity and Shannon index), while Treatment NPKM was 7.0%~66.9% higher. Besides, Treatment NPK was 10.3% and 13.0% lower, respectively in phylogenetic diversity and Shannon index; (4) Stepwise regression analysis shows that soil pH was the primary factor determining relative abundance of the dominant soil bacterial phyla and four diversity indices of the soil bacterial community. Furthermore, multivariate regression tree (MRT) analysis shows that soil pH explained 83.1% of the detected variation of the soil bacterial communities between the treatments, and differentiation of the bacterial community was driven by soil pH; and (5) Compared with CK, statistical analysis of metagenomic profiles (STAMP) shows that Treatment N, NPK and NPKM had 11, 14 and 8 treatment-specific bacterial genera, respectively.【Conclusion】In conclusion, the bacterial community in the upland red soil following long-term fertilization is mainly affected by soil pH, and the negative effects of soil acidification induced by long-term fertilization may have far exceeded the positive effects of fertility improvement. Therefore, fertilization in upland red soil should be based on the premise of prevention of soil acidification, and combined application of chemical fertilizer and manure is a suitable fertilization measure.