In order to ascertain the superior characteristics of PCBN tool for hardened material machining, to promote the cutting performance and efficiency of a mold manufacturing, to investigate the wear mechanism of the cutting tool, and to investigate the dimensional accuracy and surface finish of the machined molds, SKD11 die steel and the polycrystalline cubic boron nitride are used as the workpiece and tool materials, respectively, in this study for turning experiments. After some proper surface layers removed from the workpiece in the experiment, the tool wear was measured through the toolmaker’s microscope and the roughness of the machined surface was measured by the roughness measuring instruments. So that, the associated sampling data prepared for training pattern of a neural network can be obtained. Besides, the noise-mediator was used to detect cutting noise during each surface layer removal for the cutting performance judgment in the machining processes additionally. An assessment model of cutting process is thus developed using a neural network system if the reliable and sufficient data is taken from the experiments. Based on the developed neural network, the complicated relationships between the cutting parameters (cutting speed, depth of cut and feed rate) and the cutting performance (surface roughness, tool wear and cutting temperature) can be clearly clarified. The best surface roughness of 0.29μm Ra is obtained from these experiments under the cutting conditions of d =0.2mm, f =0.05mm/rev and V=120m/min. This surface quality is equivalent to the manufacturing process of chemical-mechanical polishing (CMP), and the surface roughness of 0.2~0.5μm Ra may be attained by CMP. The CMP is always applied to high precision surface processing such as the valve piping and connector components in semiconductor/LED manufacturing.