The assessment of credit loan application is usually carried out by loan officers based on their own heuristic judgment. Thus, different officers may have different decisions for the same application. In order to improve the assessment objective, quantitative evaluation methods have been proposed. Statistical methods, Neural Networks, Genetic Algorithms, and other forecasting methods have been used for this purpose. The present paper proposes a new Hybrid Intelligent Decision Support System (HIDSS) for credit risk evaluation, based on neural networks and genetic algorithms. The major advantages of the proposed system are higher precision in credit evaluation of the high risk customers and higher sensitivity in the evaluation of higher value loans. The proposed system is applied on a real case study concerning loan risk evaluation by a leading branch of Mellat Bank (Iran). Results are compared to the result of other forecasting methods such as statistical method and neural network.