This paper proposes a supervised training-test method with Genetic Programming (GP) for patt classification. Compared and contrasted with traditional methods with regard to deterministic patt classifiers, this method is true for both linear separable problems and linear non-separable problems. For specific training samples, it can formulate the expression of discriminate function well without any prior knowledge. At last, an experiment is conducted, and the result reveals that this system is effective and practical.