Abstract: An immune genetic algorithm is proposed based on the combination of a preference weight coding and a three-individual-crossover.This coding method not only meets the demand of operational preferential constraint in Job-shop Scheduling Problem(JSP) but also avoids unfeasible solutions effectively.The crossover operators of three-individual-crossover inherit to a great degree the excellent characteristics from their parents while guaranteeing the diversity of their descendants.The immune operator based on orthogonal experiment diversifies the antibodies,which greatly improves the algorithm's ability to overcome partial convergence.This paper uses orthogonal experiment method to confirm the parameters,which enhances the convergence speed.Simulations are made on several benchmark problems in JSP,and comparisons are made between the presented algorithm and other algorithms.With satisfactory results,the experiments prove that the proposed algorithm is valid for JSP problems.