Starting from the definition of mechanical work for an ideal gas, we present a novel derivation linking global wind power to measurable atmospheric parameters. The resulting expression distinguishes three components: the kinetic power associated with horizontal motion, the kinetic power associated with vertical motion and the gravitational power of precipitation. We discuss the caveats associated with integration of material derivatives in the presence of phase transitions and how these affect published analyses of global atmospheric power. Using the MERRA database for the years 2009–2015 (three hourly data on the 1.25° x 1.25° grid at 42 pressure levels) we estimate total atmospheric power at 3.1 W m^−2 and kinetic power at 2.6 W m^−2. The difference between the two (0.5 W m^−2) is about half the independently estimated gravitational power of precipitation (1 W m^−2). We explain how this discrepancy arises from the limited spatial and temporal resolution of the database. Our analysis suggests that the total atmospheric power calculated with a spatial resolution of the order of one kilometer (thus capturing the small moist convective eddies) should be around 5 W m^−2. We discuss the physical constraints on global atmospheric power and how considering the dynamic effects of water vapor condensation offers new opportunities.