There are two reasons making the existing intercell scheduling algorithms hardly be put into use in practice. One is the negligence of transportation among cells that affects the accuracy of the scheduling solutions, and the other one is the large problem size that decrease the computation efficiency. A novel grammatical evolution (NGE) algorithm is proposed to address the intercell scheduling problem with limited transportation capacity. This problem is divided into two subproblems: sequencing for parts and routing for vehicles. A two-stage evolutionary is developed to evolve better heuristic rules. In the first stage, string-form individuals are evolved, and in the second one, tree-form individuals are evolved. Through the comparative experiments, the effectiveness and efficiency of NGE are verified.