It is important in surface machining to estimate tool wear because it has a large effect on precision and surface roughness of a workpiece. In this study, tool wear of a ball end mill is investigated in order to design an effective estimator of the wear. Wear distribution of the tip is calculated based on the geometric model of the tool and milling parameters measured in preparatory experiments under some limited cutting conditions. The proposed algorithm of wear estimation is tailored to cutting of a plane and a curved surface. The algorithm is applied to compensate processing error due to tool wear in the case of milling of the curved surface of S55C, resulting effective reduction of the processing depth error compared with no compensation.