Plastic injection moulding is known as a manufacturing process, where material is injected into a mould and parts can be produced. The produced part’s geometry acquires the form of the respective mould, therefore the quality of the moulded products depends on how good the surface quality and shape deviations of the mould is. Nowadays moulds for plastic injection moulding are still polished by hand (Boerret et al., 2008) which is why polishing is responsible for up to 30% of the total mould price. The robot polishing process can now be used for the polishing step of the production of steel moulds. This process brings more stability to the moulds production than the manual polishing. Despite the stability that this new process brings to the production of moulds there are still some unknown factors. The current robot polishing process consists of 8 to 14 different polishing parameters. Changing these parameters result in different surface roughness and in different material removal rates. The goal of this work is to explain the robot polishing process by matching simulation and experiment allowing the prediction of the material removal. Simulating the polishing process will assure the right correction of geometric forms guaranteeing the right surface roughness and shape deviation of steel moulds. The experimental tests were performed on a robot polisher with an attached polishing head. The results of the tests were measured with the metrology equipment available in the Aalen University in the Centre for Optical Technologies: a tactile measuring machine, a white light micro interferometer and an interferometer. For the evaluation of the results the proprietary software Zaphod was used. This research has contributed to new knowledge and prediction of the material removal during the robot polishing process of steel samples. The developed simulation model in conjunction with the current polishing process enabled the improvement of the geometrical shape of hardened steel samples of more than 50%.