Despite recent advances in automation technology, geometrically complex workpieces are still often finished manually. An example for this is the post-processing of dental prostheses, which display exceptionally high surface requirements. Aiming for an improved quality assurance, this article presents a methodology for an adaptive tool path planning for polishing of geometrically complex workpieces. For this purpose, the initial roughness of the workpiece is determined using a machine-integrated measuring system. Next, suitable process parameters are selected based on machine knowledge and the adapted NC code of the polishing process is generated and simulated. The results of the simulation and the actual polishing process are compared afterwards and transformed into process knowledge. Thus, the adaption of the process parameters and the quality of the simulation are continuously improved. The article highlights the implementation of the methodology with special emphasis on the selection of the process parameters.