Abstract: In order to predict the spoilage of microorganism in frozen fish, it is important to establish the growth model of specific spoilage organisms (SSO). The traditional growth kinetics model decreased accuracy when dealing with default data and has deviation at stable stag of microbial growth. In this paper, CHAID (chi-square automatic interaction detection) method was used to predict SSO growth of large yellow croaker compared with traditional growth kinetics model. The results showed that Vibrio was SSO of large yellow croaker. The accuracy of predicted data changed with the data quantity and the data change rate. CHAID method was less than growth kinetics model at exponential stage, but was more accurate at stable stage and large quantity of data. So modeling for growth of specific spoilage organisms in chilled fish based on CHAID can be used to predict the spoilage of microorganism.