This paper firstly analyzes the significance of testing GA-hardness in the study of genetic algorithms. Be2 cause the study in this aspect nowfocuses on binary encoding , we analyze how to test GA-hardness for real encoding in genetic algorithms. After discussing the method of FDC(fitness distance correlation) and the method of correlation functions based on the randomwalk model , we propose the best one-order function approximation method to test GAhardness with real encoding. At last , a lot of experiments are done , and these methods are compared and evaluated according to the experimental results