Abstract: With the combination of artificial neural network, genetic algorithm, and immune system theory, the immune modulated symbiontic evolution （IMSE） method in neural network design is proposed. By evolving a population of neurons instead of neural networks, great computation pressure is alleviated. At the same time by utilizing the adjustment of antibody chrome and the maintenance of individual diversity, immune modulation is used effectively in preventing premature convergence and promoting population diversity. This method is verified in a pattern discrimination learning task. The experimental result shows IMSE can be used to design robust and efficient neural network.