Credit scoring model is a popular tool for the financial institutions (FIs) to assess their customers’ credit risk. Since the large amount of money in credit granting business for FIs, an improvement in the accuracy of the credit scoring model to recognize good and bad customers, even a fraction of one percent can help to reduce significant loss. Some existing researches suggest that adaboost model can help to improve the accuracy of classification for base classifiers. In this paper, two adaboost models with different weights strategies are introduced for credit scoring. Multilayer perceptron neural network with back-propagation training method is employed as the base classifier. The models are tested on one real-world dataset and the experimental results show that adaoosting neural network model is outperformed than the single neural network and traditional adaboost model.