One of the primary reasons water-quality standards are not met is the effect of storm runoff and combined sewer overflows. A methodology is presented here to determine the duration of storm effects on stream water quality. The evaluation of the duration of storm effects on water quality involves two steps. First, calibration of an appropriate water quality model that is capable of simulation of unsteady-state conditions. Second, execution of the calibrated model with a number of storm loadings randomly sampled from a specific probability distribution that represents realistic ranges of pollutant concentrations. When the variations in the simulated water quality variables become negligible, it is assumed that the river system goes back to prestorm, dry-weather conditions. To illustrate this methodology, the DUFLOW unsteady-state water quality model and Latin hypercube sampling are applied to evaluate the duration of storm effects on water quality in the Chicago Waterway System (CWS). The duration of the storm impacts on dissolved oxygen lasts 2 days-2 weeks in the CWS depending on the location in the system and the magnitude of the storm. Moreover, a strong relation between the precipitation depth and the duration of the storm effects on in-stream water quality constituents was found in the CWS. Outcomes of this research suggest that the duration of the storm effect on water quality can be reasonably predicted with the help of robust unsteady-state water quality models.