The authors propose an approach towards detecting credit card frauds using rough set theory and artificial neural networks. They aim at finding an optimal attribute reduct using rough set theory and then using these attributes to train an ANN system which would then carry out the credit card fraud detection. This hybrid model will maximize the correct diagnosis by minimizing both the number of false alarms and the number of fraud transactions not recognized. The foremost step in their proposed model is the reduction of the transaction data set to ensure a manifold decrease in space and time requirement for fraud detection. This pre-processed data will then be used for learning and validating the system.