The quality of the surface texture on cylinder liners of engines has an important influence on oil consumption and running performance. This texture is generated with a succession of three honing operations (rough, finishing and plateau honing). These honing steps are difficult to optimize due to the large number of process variables (Velocity of the rotational motion, velocity of the axial motion ...) and their complex interrelation. This paper discusses the use of the multiscale topographic parameter of the honing process as a new concept to identify the relation between the process variables and the resulting performances in terms of form and surface finish. To achieve this objective, a multi-scale decomposition method of surface topography based on 2D continuous wavelets transform was performed to analyze the evolution of the topographic honing. This approach also shows a good indication of the dependence between the honing steps and the roughness scales. Finally, it provides an efficient tool to predict the surface characteristics corresponding to given process variables and can be extended to optimize other multi-stage abrasive machining processes.