The goal of this study is to present a completely new procedure for automatic occlusal surface reconstruction and to test it in several inlay situations. The starting point is the mathematical analysis of a large number of occlusal surfaces from a tooth library. Learning algorithms used in pattern recognition and image processing were adapted for the special needs of dental morphology. A biogeneric tooth model resulted which can mathematically describeacertain type of tooth only with a few parameters. This model was tested on 40 simulated inlay situations and as an example on two clinical inlay situations. The results show that fully automatic reconstruction was possible except for four cases. In a majority of the cases, no further interactive improvements would have been necessary. Because of the general, theoretically wellfounded approach, fully automatic reconstruction of the occlusal surface can also be extended in the future to full crowns or reduced crown substructures (copings). Keywords: CAD, occlusal, occlusal surface reconstruction, biogeneric, tooth model, inlay, onlay