In the Soil Conservation Service Curve Number (SCS-CN) method, the three levels of antecedent moisture condition (AMC) permit unreasonable sudden jumps in curve numbers, which result into corresponding jumps in the estimated runoff. A few recently developed SCS-CN-based models obviate this problem, yet they have several limitations. In this study, such a model incorporating a continuous function for antecedent moisture has been presented. It has several advantages over the other existing SCS-CN-based models. Its application to a large dataset from US watersheds showed to perform better than the existing SCS-CN method and the others based on it. Copyright © 2010 John Wiley & Sons, Ltd.