The buoyancy balance between two outcropping isopycnals leads to diagnostics for quantifying water mass formation rates between them due to air-sea buoyancy fluxes <Walin, 1982>. The surface air-sea transformation gets modified by mixed layer processes so that the net formation rate below the winter mixed layer depth is different from that given by surface air-sea fluxes alone. Here we estimate the role of time dependence and mixed layer deepening to quantify the water mass transformation due to mixed layer entrainment fluxes. We focus on the mixed layer transformation in the North Atlantic for 1990--2000 during the World Ocean Circulation Experiment period, using both entrainment parameters and isopycnal geometry. The water mass transformation due to mixed layer entrainment is calculated using a large number of gridded one-dimensional mixed layer models forced by National Centers for Environmental Prediction (NCEP) reanalysis air-sea fluxes (daily/6 hourly) to calculate the local entrainment parameters; Reynolds sea surface temperature and Levitus monthly salinity data determine the isopycnal geometry. To get a closed annual cycle in the mixed layer depth, any net annual heat flux is ascribed to advective processes. These are included in the annual mean, leaving any synoptic forcing in the NCEP forcing unperturbed. In general, the mixed layer transformation opposes air-sea interaction, with amplitudes of (1.3 Sv) in the equatorial region (without equatorial upwelling contribution) and (0.5 Sv) in the overflow region. Mean cross-isopycnal volume fluxes are O(1 Sv), with considerable interannual variability. These estimates of water mass formation due to mixed layer processes are sensitive to synoptic frequencies, but not to climatological mean air-sea fluxes, and are within the imposed noise levels for inverse box models of the North Atlantic.