The main goal of this work is to show the feasibility of using neural network in tool vibration monitoring systems aiming to automatically establish the end of turning tool life. For this purpose, a complete vibration monitoring system was built, with two accelerometers attached to the tool holder and with the vibration signals being stored in a PC computer. Several turning experiments were carried out to cut the AISI 4340 steel with different feeds and various cutting speeds which are typical of finish turning operations. The tool vibration signals were acquired and stored in the computer memory and the surface roughness of the parts was measured after the cuts. After that, a back propagation neural network was run. The input parameters were feed, cutting speed, cutting length and tool vibration signals. The output parameter was the status of the tool. The most important conclusion of this work is that the neural network is able to establish an interval where the tool replacement must be done. In this interval the tool is already worn, but not completely deteriorated The part surface roughness is close to its limit in finish turning operations, but the part is still acceptable.