To examine the diversity and function of the bacterial community in the reservoirs at the middle part of Northern Tianshan Mountain, Moguhu reservoirs (MGW), Kuitun/Chepaizi reservoirs (KCW), Anjihai reservoirs (AJW), and Bayi reservoirs (BYW) were selected as the typical reservoirs for metagenomics analysis. The surface water samples of the reservoirs were collected at multiple points, the bacteria in the water samples were collected by centrifugation. The bacterial DNAs were extracted by CTAB method, the V3-V4 regions were amplified with bacterial 16S rDNA universal primer followed by high-throughput sequencing. The software and online tools including BLAST, USEARCH and QIIME were used to analyze the bacterial diversity, and the functional gene compositions were predicted using PICRUSt software and KEGG database. The results showed that the bacteria from the four reservoirs belonged to 18 phyla, 38 classes and 181 genera, the operational taxonomic units (OTUs) number and diversity index of AJW was higher. There was something in common in the bacterial community compositions of the four reservoirs. The Proteobacteria and Bacteroidetes were the dominant bacteria at the phylum level, Flavobacteria was the dominant bacteria at the class level, and Flavobacterium was the dominant bacteria at the genus level. However, there were great differences in the bacterial community compositions of the four reservoirs. In Proteobacteria, Gammaproteobacteria showed a higher relative abundance in MGW and BYW, and Alphaproteobacterial and Betaproteobacteria showed a higher relative abundance in AJW and KCW, respectively; Bacillus exhibited a higher relative abundance in AJW and MGW; other bacteria with higher abundance included Planomicrobium in AJW, Massilia in KCW, Acinetobacter and Planomicrobium in MGW, and Pseudomonas and Psychrobacter in BYW; in addition, some specific high-abundance classes and genera were found in AJW, among them, both Exiguobacterium and CL500-29_marine_group were conducive to maintaining the water quality and bacterial diversity. PICRUSt function prediction analysis indicated that the genes of the bacterial community in four reservoirs related to metabolism, genetic information processing, environmental information processing and cellular processes showed relatively a high abundance. In addition, some bacterial genes were related to the biodegradation of xenobiotic compounds, and the relative abundances of genes related to degradation of benzoate, aminobenzoate, chloroalkane and chloroalkene, and naphthalene were higher. In summary, the compositions of the bacterial community in different reservoirs were varied, which could be related to the water source, surrounding environment, and human activities. The presence of the genes related to the degradation of aromatic compounds and organochlorine compounds in the reservoirs could play a potential role in pollutant degradation and bioremediation.