This study aimed to provide better understanding of the bedrock aquifer bacterial communities and their functions in deep geological repository (DGR) environment. Two study sites of uranium deposits in the Ogcheon Metamorphic Belt were selected: Boeun and Guemsan. From two study sites, six groundwater samples were obtained with different boreholes and depths: OB1 (Boeun, 25 m), OB3 (Boeun, 80 m), GS1 (Guemsan, 25 m), GS2 (Guemsan, 85-90 m), GS3-I (Guemsan, 32- 38 m), GS3-II (Guemsan, 70-74 m). The physicochemical properties of groundwater were analyzed by multi-parameter sensors, ion chromatography (IC), and inductively coupled plasma optical emission spectroscopy (ICP-OES). Illumina Miseq sequencing was performed to investigate bacterial community in six groundwater samples. In addition, the number of sulfate-reducing bacteria (SRB) was quantified by a quantitative PCR (qPCR). Bacterial community composition varied in response to boreholes and depths. A total of 14 different phyla and 36 classes were detected from six groundwater samples. Overall, Proteobacteria, Actinomycetota, and Bacteroidota were dominant in the phylum level. SRB and iron-reducing bacteria (IRB) were detected in all groundwater samples even though organic carbon sources were not abundant (0.7-3.3 mg-total organic carbon/L). This result shows a potential to immobilize uranium in DGR environment. In particular, SRB, Desulfosporosinus fructosivorans and Humidesulfovibrio mexicanus were mainly detected in GS1 and GS2 groundwater samples, which attributed to higher dissimilatory sulfite reductase functional gene copy number in GS1 and GS2 groundwater samples. Statistical analysis was performed to understand the correlation between environmental factors and core bacterial species. Dissolved oxygen (DO), Fe, and Mn concentrations were positively correlated with Curvibacter fontanus while Undibacterium rivi had a negative correlation with pH. These results indicate that bacterial community could be changed in response to environmental variation. Further study with a greater number of samples is necessary to obtain statistically reliable and meaningful results for a safe DGR system.