This paper addresses the possible role of unresolved terrain drag, relative to the turbulent drag on the development of the stable atmospheric boundary layer over land. Adding a first-order estimate for terrain drag to the turbulent drag appears to provide drag that is similar to the enhanced turbulent drag obtained with the so-called long-tail mixing functions. These functions are currently used in many operational models for weather and climate, although they lack a clear physical basis. Consequently, a simple and practical quasi-empirical parameterization of terrain drag divergence for use in large-scale models is proposed and is tested in a column mode. As an outcome, the cross-isobaric mass flow (a measure for cyclone filling) with the new scheme, using realistic turbulent drag, appears to be equal to what is found with the unphysical long-tail scheme. At the same time, the new scheme produces a much more realistic less-deep boundary layer than is obtained by using the long-tail mixing function.