In the multiple cars elevator (MCE), multiple cars move in the shafts. In MCE, a special control algorithm is necessary for avoiding the conflict between cars. We have presented the group control algorithm with non-overlapping service zones for MCE. In this algorithm, each car has the non-overlapping operating zone and therefore, the conflict between cars can be avoided perfectly. On the other hand, its performance depends on the initial service zones. In this study, genetic algorithm is applied to the optimization of the service zones of cars. The numerical results show that the average operation time of the group control algorithm with optimized service zones is smaller than that of the non-optimized service zones.