In this research, the data of customer's basic informations and accounts receivable status from an electronic company in 2007, a total sample of 224 customers, were collected. First, the customers were grouped by account receivable status with cluster analysis. The characteristics of each group were identified with basic statistical description, the chi-square test of independence and ANOVA method. Second, the account overdue prediction model was established by logistic regression method that provided small and medium enterprises the use of credit in the financial management.