Deriving overland evapotranspiration (ET) estimates is an important part of the larger effort to develop long-term Earth System Data Records (ESDRs) for the major components (storages and fluxes) of the terrestrial water cycle. In the current study, global estimates of sensible heat and evaporative fluxes are developed for 1984-2006 using three process-based models forced by two remote sensing based data sets. The models are surface energy balance system (SEBS), a modified Penman-Monteith approach, and a Priestley-Taylor approach. The models are driven by radiation inputs from the ISCCP and SRB data sets, with the meteorological forcing data from ISCCP, and vegetation characteristics from AVHRR. Estimates are made using the three models. Comparisons among the data sets show large differences in magnitude and long-term variability, due mainly to uncertainties in the forcing radiation. Comparisons with independent data sets from inferred evaporation estimates [(P-Q)climatology], off-line land surface model (VIC) data, previously developed remote sensing products and estimates derived from tower data, reveals consistency at large scales, but large differences in some regions, most notably in the northern hemisphere.