The U.S. Geological Survey (USGS), in cooperation with the Minnesota Department of Natural Resources and the Heron Lake Watershed District, conducted a study to characterize the rainfall-runoff response and to examine the effects of wetland restoration on the rainfall-runoff response within the Heron Lake Basin in southwestern Minnesota. About 93 percent of the land cover in the Heron Lake Basin consists of agricultural lands, consisting almost entirely of row crops, with less than one percent consisting of wetlands. The Hydrological Simulation Program – Fortran (HSPF), Version 10, was calibrated to continuous discharge data and used to characterize rainfall-runoff responses in the Heron Lake Basin between May 1991 and August 1997. Simulation of the Heron Lake Basin was done as a two-step process: (1) simulations of five small subbasins using data from August 1995 through August 1997, and (2) simulations of the two large basins, Jack and Okabena Creek Basins, using data from May 1991 through September 1996. Simulations of the five small subbasins was done to determine basin parameters for the land segments and assess rainfall-runoff response variability in the basin. Simulations of the two larger basins were done to verify the basin parameters and assess rainfall-runoff responses over a larger area and for a longer time period. Best-fit calibrations of the five subbasin simulations indicate that the rainfall-runoff response is uniform throughout the Heron Lake Basin, and 48 percent of the total rainfall for storms becomes direct (surface and interflow) runoff. Rainfall-runoff response variations result from variations in the distribution, intensity, timing, and duration of rainfall; soil moisture; evapotranspiration rates; and the presence of lakes in the basin. In the spring, the amount and distribution of rainfall tends to govern the runoff response. High evapotranspiration rates in the summer result in a depletion of moisture from the soils, substantially affecting the rainfall-runoff relation. Five wetland restoration simulations were run for each of five subbasins using data from August 1995 through August 1997, and for the two larger basins, Jack and Okabena Creek Basins, using data from May 1991 through September 1996. Results from linear regression analysis of total simulated direct runoff and total rainfall data for simulated storms in the wetland-restoration simulations indicate that the portion of total rainfall that becomes runoff will be reduced by 46 percent if 45 percent of current cropland is converted to wetland. The addition of wetlands reduced peak runoff in most of the simulations, but the reduction varied with antecedent soil moisture, the magnitude of the peak flow, and the presence of current wetlands and lakes. Reductions in the simulated total and peak runoff from the Jack Creek Basin for most of the simulated storms were greatest when additional wetlands were simulated in the North Branch Jack Creek or the Upper Jack Creek Subbasins. In the Okabena Creek Basin, reductions in simulated peak runoff for most of the storms were greatest when additional wetlands were simulated in the Lower Okabena Creek Subbasin.