Abstract Comprehensive boundary layer measurements from a drift station on first-year ice in the late summer of 2012 in the Nansen basin, when stable stratification in the upper ocean extended all the way to the surface, are analyzed. Observed quadratic ice–ocean drag coefficients, based on measurements of wind stress, are roughly 3.6 × 10−3, consistent with neutral-stability Rossby similarity scaling. The turning angles of 32°–39° between surface velocity and stress are larger than Rossby similarity predicts and obey a different scaling. This can be explained by the shallow pycnocline forcing the Ekman transport into a thin layer and modeled roughly employing a simple first-order correction to Rossby similarity. Turbulent shear stress in the ice–ocean boundary layer is on average 3 times smaller than the estimate based on wind stress, possibly because internal wave drag was significant. This lowers vertical scalar fluxes by 38% compared to a scenario where turbulent stress accounts for the total drag. The authors measure an average upward ocean–ice heat flux of 10 W m−2, which is 50% smaller than predicted by a bulk heat flux parameterization. This reduction is attributed to additional sources of heat and freshwater that alter the ice–ocean interface salt balance. This study shows that a commonly used bulk heat flux parameterization is a special case of a simple downgradient parameterization allowing for a modified interface salt budget. For similar wind forcing, observed ice–ocean fluxes of heat and salt were 40%–100% larger when the ice-relative current approached from a nearby pressure ridge keel than otherwise.