In this paper，a genetic classifier system for complex system analysis is proposed. The system not only makes use of the remarkable genetic algorithms searching ability and the ability to find laws from a great amount of historical data to acquire classifiers，but also posseses the following two charateristics: ①It is suitable for such complex systems which have noise and lack complete description ②The credit of each classifcer acquired by this system is identical to statistical results，so it can support meta synthesis reasoning method.