The quality of the customer service has become a point of competitive distinction and positional advantage. The functionality of remote system diagnosis and troubleshooting can increase the quality of customer service. However, to provide such a new functionality, business must possess a highly concentrated backstage support capability. In this study we develop an intelligent self customer trouble-shooting assistant system using a hybrid method which integrates Bayesian Belief Network (BBN) and Case-based Reasoning model for customer problems solving. The proposed method performs probability inference to model uncertain domain knowledge and allows unique experience to be memorized and retrieved in a more easy way. The techniques described are demonstrated by an example developed in our laboratory. The example can be a reference model for the people who are interesting to develop a self customer service system.