The development of algorithms for the retrieval of water cycle components from satellite data - such as total column water vapor content (TCWV), precipitation (P), latent heat flux, and evaporation (E) - has seen much progress in the past 3 decades. In the present study, we compare six recent satellite-based retrieval algorithms and ERA5 (the European Centre for Medium-Range Weather Forecasts' fifth reanalysis) freshwater flux (E P) data regarding global and regional, seasonal and interannual variation to assess the degree of correspondence among them. The compared data sets are recent, freely available, and documented climate data records (CDRs), developed with a focus on stability and homogeneity of the time series, as opposed to instantaneous accuracy. One main finding of our study is the agreement of global ocean means of all E P data sets within the uncertainty ranges of satellite-based data. Regionally, however, significant differences are found among the satellite data and with ERA5. Regression analyses of regional monthly means of E, P, and E P against the statistical median of the satellite data ensemble (SEM) show that, despite substantial differences in global E patterns, deviations among E P data are dominated by differences in P throughout the globe. E P differences among data sets are spatially inhomogeneous. We observe that for ERA5 long-term global EP is very close to 0mmd1 and that there is good agreement between land and ocean mean E P, vertically integrated moisture flux divergence (VIMD), and global TCWV tendency. The fact that E and P are balanced globally provides an opportunity to investigate the consistency between E and P data sets. Over ocean, P (nearly) balances with E if the net transport of water vapor from ocean to land (approximated by overocean VIMD, i.e., r - .vq/ocean) is taken into account. On a monthly timescale, linear regression of Eoceanr - .vq/ocean with Pocean yields R2 D 0:86 for ERA5, but smaller R2 values are found for satellite data sets. Global yearly climatological totals of water cycle components (E, P, EP, and net transport from ocean to land and vice versa) calculated from the data sets used in this study are in agreement with previous studies, with ERA5 E and P occupying the upper part of the range. Over ocean, both the spread among satellite-based E and the difference between two satellite-based P data sets are greater than E P, and these remain the largest sources of uncertainty within the observed global water budget. We conclude that, for a better understanding of the global water budget, the quality of E and P data sets needs to be improved, and the uncertainties more rigorously quantified.