WHEN the problem of drag computation from computational uid dynamics calculations was addressed, Slooff1 asked,“Mission impossible?” Despite the progressin this fi eld, the question is still open today, and much research has been devoted to this subject. In addition to the dedicated conference where Ref. 1 was presented, it is possible to fi nd a review with the fundamentals of physics in Ref. 2 and a more recent, extended, and detailed overview of the state of the art on drag prediction methods in Ref. 3. The numerical computation of drag by surface integration of stresses (near-fi eld method) usually gives insuffi ciently accurate results even if the ow solution is locally accurate (in terms of pressure and velocity profi les, for instance). In particular, for numerical solutions of the Euler/Reynolds averaged Navier–Stokes (RANS) equations, which are discussed in this work, the problem is mainly related to the presence of the numerical artificial dissipation and of the discretization error, which produces an artifi cial or spurious drag. This contribution becomes negligible only for unfeasible calculations with infi nitely dense grids. A second problem is that the near-fi eld drag computation only allows for a distinction between pressure and friction drag. Additional useful information would be the breakdown into other physical components, such as viscous drag (associated with boundary layers), wave drag (associated with possible shock waves in transonic and supersonic ows), and lift-induced or vortex drag (associated with the free-vortex system shedding from three-dimensional lifting bodies). This task is relatively simple when drag has to be extracted by classical viscous–inviscid interaction methods. However, in the case of analysis performed by RANS methods such as in wind-tunnel experiments, the physical drag source is not isolated, and the breakdown into individual components becomes diffi cult. In practice, for a real ow there is not a clear defi nition of the different drag contributions.