During the last decade the methods for solving optimal scheduling problems have been improved. But it is still hard to find out the optimal or very near optimal solution for large size batch process scheduling problems. Ku and Karimi (1991) developed a simulated annealing (SA) method for solving scheduling problems and showed that SA offers good performance but the control parameters of SA must be tuned when the problem constraints are changed. In this work, we develop a genetic algorithm (GA) for effectively solving large-size scheduling problems. The application of GA to multi-product batch process scheduling problems with several intermediate storage policies is treated. Particular form of GA is shown to be suitable for this class and scheduling problems without tuning of algorithm parameters for different problem parameter sets. We solved various size of problems for the minimization of makespan with unlimited intermediate storage (UIS) and zero wait (ZW) storage policies to test the performance of GA. GA is shown to be superior to heuristic of SA-based search methods. (C) 1998 Elsevier Science Ltd. All rights reserved.