The roughness of polymer surfaces is often investigated to guarantee both the surface integrity and the surface functionality. One of the major problems in roughness measurement analyses consists in determining both the evaluation length and the reference line (i.e., the degree of the polynomial equation) from which roughness parameters are computed. This article outlines an original generic method based on the generalized analysis of variance and experimental design methodology for estimating the most relevant roughness parameter p, the most pertinent scale, s, and finally, the degree of the polynomial fitting, d. This methodology is then applied to characterize the influence of four process parameters on the final roughness of poly(polypropylene) samples obtained by injection molding. This method allows us to determine the most efficient triplet (p, s, d) that best discriminates the effect of a process parameter q. It is shown that different (p, s, d) values are affected to each process parameter giving finally the scale on which each process parameter modifies the roughness of a polymeric surface obtained by injection molding. POLYM. ENG. SCI., 2008. © 2008 Society of Plastics Engineers