On the basis of analyzing and studying the correlation technique of Pathology Expert System on Artificial Neural Network, the paper elaborates the Inference Mechanism based on artificial neural networks and expert system in detail. With the knowledge and experience of pathology experts, we design an expert system of recognizing automatically abnormal cell by using technology of pattern recognition. We do the simulation experiments of this expert system on the Visual C++ 6.0 platform. Using a two-tier network to achieve it, BP network has the classification of the good performance and the shorter training time and LVQ network has the credibility of the classifier to promptly adjust the corresponding BP sub-classifier. The experimental result shows that this expert system has ability to carry out pathological diagnosis of unconventionality cell, and the classification accuracy of unconventionality cell achieves 93.71%.