Theory and experiments show that a component of compressive strain parallel to layering is necessary to develop drag folds. The component of shear strain needed for the monoclinic symmetry of such folds is not sufficient to form the wavy pattern. Based upon fluid dynamic theory and experiments with layered rubber slabs it is suggested that drag folds on flanks of major folds generally evolve as follows: Layer-parallel compression of more or less straight layered rocks causes some layers to shorten essentially by more or less uniform compressive strain and thickening (thick and not-so-competent layers), other layers to shorten essentially by buckling (thin and most component layers). In the course of time, however, the rate of buckling of the thick layers becomes significant relative to the rate of arc-shortening, and the thin layers with the early formed short buckles are forced to mimic the larger folds of the thick layers more or less passively. In this process the early formed short buckles become tilted and assume the typical drag-fold shape.