A mesoscale numerical model is used to compute the different components of the pressure drag on mountains, i.e.: form drag, wave drag, hydrostatic drag, and total pressure drag, for stable stratification. The Froude number is chosen so that non-breaking lee-waves evolve. The paper explains how the different parts of the drag are computed from the numerical results and how they form part of the horizontal momentum budget. For a single mountain the drag from the evolving stationary solution is compared to the wave drag from linear inviscid theory. Wave drag turns out to be about one third of the value expected from linear theory, and nonlinear interaction between wave and form drag is found. Wave drag is responsible for about 75% of the total drag if blocking is negligible. For two obstacles with varying distance the wave drag in the stationary solution varies between 5% and 30% of the value from linear theory due to partial cancellation between the lee-waves from the two mountains. Finally in an instationary simulation the passage of a cold air mass over a mountain and the respective drag components have been computed. 1500 m above the crestline of the obstacle wave drag is only 10% to 30% of the total drag. From the present results it seems realistic that wave drag from ALPEX experimental data was only a few percent of the value expected from the surface pressure distribution and from linear theory.