A method of artificial neural network was presented to forecast the compressive strength of clay brick in view of the questions existing in the ultrasonic and rebound strength testing method. A relative forecasting model was established base on the principles of the improved BP neural network. The neural network was trained by the test data, and a case study of a real engineering was tested. The results show that the forecasting results obtained from the neutral network method is better than that from the traditional mathematic model.