Recursive Partitioning Algorithm (RPA) is introduced as a technique for credit scoring analysis, which allows direct incorporation of misclassification costs. This study corroborates nonagricultural credit studies, which indicate that RPA outperforms logistic regression based on within-sample observations. However, validation based on more appropriate out-of-sample observations indicates that logistic regression is superior under some conditions. Incorporation of misclassification costs can influence the creditworthiness decision.