In order to overcome the weakness of redundant data, poor knowledge, cutting parameters being not closely integrated with the real cutting conditions in traditional metal cutting databases,an expert system with self-learning function for turning using carbide turning tools was presented. In this system, the cutting parameters were stored by using the concentrated mathematical models obtained from the cutting experiment, the cutting mathematical models were amended by the self-learning function, thus the reasonable and optimal turning tools, cutting speed as well as the life of turning tools etc, can be recommended, and the processing quality and moved metal can be predicted. A special turning system with its own feature will be formed.