This study proposed a modified nonlinear storage function runoff model to take into account the storage hysteresis effect, in which there exists difference of the storage-discharge relationship between the rising and recession limb. Since the modified storage function runoff has seven parameter, a parameter-calibration method, which combines the genetic algorithm with the least square criterion. For model calibration and validation, twenty rainfall-runoff events (1968-2005) recorded at Wudu gage in Keelung River in northern Taiwan were used in the study. The results of model validation reveal that the modified storage function runoff model not only produces the realistic storage-discharge relationship, but also provides a good estimation of the runoff. (C) 2011 International Association of Hydro-environment Engineering and Research, Asia Pacific Division. Published by Elsevier B.V. All rights reserved.