Assessing changes in extreme hydroclimatic phenomena requires accurate precipitation estimates with high temporal and fine spatial resolutions. Significant differences were obtained between reanalysis data and observations at weather stations. Bias correction method applied to ERA5 total precipitation product with spatial resolution 0.25×0.25 ° is reported in the present paper to improve the reproducibility of the local weather conditions. Hourly ERA5 precipitations were aggregated and monthly precipitation sums were processed. Monthly precipitations for 133 weather stations in Siberia for May-September from 1979 to 2017 were used as reference data. Comparative analysis has shown that long-term average monthly precipitations from ERA5 reanalysis are higher than precipitation estimations from the observation data. The number of weather stations where the difference between two data sets exceed 60% vary from 3 in July to 37 in May. To eliminate significant differences between reanalysis and observation data, we have developed a specific procedure for correcting reanalysis data. As a result of the adjustment, the deviations of the reanalysis data from the observation data has decreased. The difference after correction is lower than 60 % at all weather stations, and from 94 (July) to 118 (August) weather stations has the difference lower than 30%.