Abstract: To evaluate the applicability of ERA5 reanalysis data in severe convective weather in Beijing during summer, ERA5 reanalysis data and soundings of Beijing at 1400 BT from June to August during 2009-2023 are used to calculate convection parameters. Differences in convective thermal parameters, water vapor, wind shear, and characteristic layer thickness between ERA5 dataset and soundings are analyzed. Results indicate that ERA5 performs well in characterizing temperature of each vertical layer, the lift index, and the lifting condensation level. However, ERA5 frequently exhibits significant deviations in convective available potential energy estimation. Most of the low and middle convective available potential energy values are overestimated, but the high convective available potential energy values above 2000 J·kg^-1 are significantly underestimated. The error in ERA5’s overestimating the medium and low convective effective potential energy values primarily stems from the overestimation of the ground temperature. The underestimation of high and convective effective potential energy values is primarily attributed to the underestimation of elevated ground dew point values. ERA5 can well describe the vertical wind shear from surface to middle troposphere (0-6 km and 0-8 km). However, there are obvious deviations in its representation of the vertical wind shear from surface to the lower troposphere (0-1 km and 0-3 km). It is more accurate in describing the dew point of lower layer, but the estimation of the water vapor condition of middle troposphere and the entire layer is excessively high (wet), resulting in a significantly lower estimate of downdraft convective available potential energy. Estimation for hail-related characteristic layer height and thickness of ERA5 is close to soundings. The analysis of three typical severe convection processes in Beijing during summer indicates that ERA5 data overestimate the water vapor conditions in both the entire layer and the middle layer. It also overestimates values of low and middle convective available potential energy, underestimating values of high convective available potential energy. However, the vertical wind shear, the characteristic layer height and thickness, are found to be in agreement with observations. The estimation increases the probability of heavy precipitation but decreases the probability of thunderstorm winds. It is challenging to provide a reliable estimate of the tornado activity based on ERA5 data, as its estimation of 0-1 km vertical wind shear exhibits significant deviations. Error sources of thermodynamic parameters are determined by the temperature profile and the dew point profile. The accuracy of the temperature profile is significantly better than that of the vertical profile of dew point temperature.