 This study evaluates 22-grade precipitation data sets for the period 2000 to 2016 using daily gauge observations from 76,086 gauges worldwide in hydrological modeling to calibrate the HBV conceptual model against stream flow records. The results show marked differences in spatial temporal patterns and accuracy among the data sets. Among the uncorrected P data sets, satellite and reanalysis-based MSWEPing version 1.2 and version 2.0 generally showed the best temporal correlations with gauge observations while estimates based primarily on thermal infrared imagery, Gridsat version 1.0, Persian, and Persian CCS performed poorly. Among the corrected P data sets, those directly incorporating daily gauge data generally provided the best calibration scores. The study highlights large differences in estimation accuracy and emphasizes the importance of P data set selection in both research and operational applications. This article was authored by H.E. Beck, and for Goplin, M-Pan, and others. We are article.tv, links in the description below.