【Objective】 To cope with the trend of global warming, CO₂ capture and storage （CCS）is one of the major technologies for reduction of CO₂ emission. And then the captured CO₂, nil in commercial value, is injected underground to raise the output of petroleum or coalbed methane, so as to maximize its economic profit. However, the prevailing technologies are far from being capable of guaranteeing zero leakage of the stored CO₂ during the processes of CO₂ storage and transportation, geological activities and human misoperation. Once the risk of potential CO₂ leakage becomes real, the leakage will pose an enormous threat to the near surface ecosystem. It is, therefore, essential to explore in depth effects of underground CO₂ leakage on farmland soil environment, especially tolerance and sensitivity of soil microbial communities to different concentrations of CO₂. So the study was conducted. 【Method】 An experimental platform to simulate underground CO₂ leakage was constructed in an idle farmland, in the South Lake Campus of the China University of Mining and Technology. The platform was used to simulate leakage of CO₂ varying in intensity, i.e. 400 g m^-2d^-1, 800 g m^-2d^-1, 1 200 g m^-2d^-1and 2 000 g m^-2d^-1 for 60 days, thus forming four treatments, i.e. L-400, M-800, H-1200 and E-2000. Besides, the experiment also had a control group and a recovery group. Soil samples were collected from the four treatments and their 3 replicates on the 14th, 30th and 60th day after the start of simulated leakage, from the control group on the day before the start of leakage (C1) and at the end of the experiment (C2), and from the recovery group, which was actually Treatment E2000, 60 days after the stop of the leakage. The soil samples were analyzed for soil physical and chemical properties, concentration of occluded soil, structure and α and β diversities of soil microbial community with the conventional physicochemical analysis method and the Illumna second generation gene sequencing method based on the Miseq platform.【Result】 Results show that CO₂ leakage decreased soil pH, electrical conductivity and nitrate nitrogen content and the effect was enhanced with rising CO₂ concentration whereas it had an opposite effect on soil organic matter content. In all the four treatments. Soil CO2 concentration increased till it reached saturation in 24 h, when soil CO₂ gas concentration leveled off at 1.60%, 4.80%, 10.80% and 19.60%, respectively. Along with increasing CO₂flux, soil microbial community decreased in diversity, Chao index and Shannon index, by 17.00%~27.80% and 6.10%~9.50%, respectively. In contrast, soil microbial community increased in β diversity (NMDS index) in Treatments L-400 and M-800 (low CO₂ concentration) but decreased in Treatments H-1200 and E-2000. Some bacteria, like Bacteroidales, varied extremely, either rising up or falling down in relative abundance with increasing CO₂ leakage. 【Conclusion】 The structure, diversity and abundance of soil microbial community varied significantly from treatment to treatment. Soil pH and CO₂ flux were the two most important environmental factors affecting soil microbial diversity. Bacteroidales was very sensitive to CO₂stress so that it can be used as a key indicator in monitoring and evaluating ecological risk of underground CO₂leakage. The 60 d short term recovery experiment indicates that the soil microbial community recovered well in diversity and richness, but it is still unclear whether it did in function. Therefore, in future studies, focuses should be laid on impacts of underground high concentration CO₂ leakage on functions of the soil ecosystem.