The lending rate and lending credit rely on the accurate evaluation of the credit risk of small business. In this paper, according to the running processes in some commercial bank in Nanjing in China, we construct an index system for small business credit evaluation which is a combination of qualitative indices and quantitative ones. Moreover we present an incremental clustering algorithm based on the chemical recognition system of ants to produce index weights. At last we adopt empirical study and the result shows that the credit value of the business gained from the model is credible and accurate.