Knowledge of the tool wear state in machining has become an important issue in research and industrial application. Current systems use the spindle power or cutting force as measured variable and refer it to a taught set point. However, this method lacks the ability to adapt to new work piece geometries. This new approach focusses on the tool instead of the workpiece, and uses a sensory tool holder with integrated strain gauges. Process monitoring that is based on the increasing effect of tool wear on the cutting force requires online information of the current engagement conditions to normalize the force information. The research shown in this paper hence presents a method to estimate the width of cut in end milling. The signal of the bending torque of the tool holder in the XY-plane shows a characteristic shape for each respective engagement condition. An analytically describable relationship has been found between these shapes for different engagement conditions. On the base of this relationship, an ANN was taught to estimate the width of cut without prior knowledge of the cutting path. The results show an accurate estimation for a wide range of cutting parameters.