The purpose of this research is to determine the feasibility of automatically generating adaptive feedrates for five-axis CNC end milling. The complicated geometries involved with multiaxis machining make it difficult to manually estimate acceptable feedrates without being overly conservative. Our strategy is to use a computer simulation of the machining process to estimate the feeds based on in-process cutting information. The simulation consists of two distinct portions: a discrete geometric model of the material removal process, and a discrete mechanistic model of the cutting forces involved. The mechanistic model estimates cutting forces as a function of material properties of the stock and cutting tool, cut geometry, and feedrate. Used in an inverse manner, the mechanistic model can estimate the feedrates necessary to maintain a constant cutting force. This force may be selected to maintain a desired part tolerance, or to meet some other criteria (e.g. machine constraints). The cut geometry information required by the inverse mechanistic model is provided by the geometric model of the material removal process. The geometric model also dynamically stores in-process stock geometry as the simulation progresses. The results of this research has shown that it is possible to automatically generate adaptive feeds using these combined models.