The U.S. Geological Survey, in cooperation with a consortium of 19 cities, towns, and villages in Dane County, Wis., undertook a study to compare the capability of rain gardens with different vegetative species and soil types to infiltrate stormwater runoff from the roof of an adjacent structure. Two rain gardens, one planted with turf grass and the other with native prairie species, were constructed side-by-side in 2003 at two locations with different dominant soil types, either sand or clay. Each rain garden was sized to a ratio of approximately 5:1 contributing area to receiving area and to a depth of 0.5 foot. Each rain garden, regardless of vegetation or soil type, was capable of storing and infiltrating most of the runoff over the 5-year study period. Both rain gardens in sand, as well as the prairie rain garden in clay, retained and infiltrated 100 percent of all precipitation and snowmelt events during water years 2004-07. The turf rain garden in clay occasionally had runoff exceed its confining boundaries, but was still able to retain 96 percent of all precipitation and snowmelt events during the same time period. Precipitation intensity and number of antecedent dry days were important variables that influenced when the storage capacity of underlying soils would become saturated, which resulted in pooled water in the rain gardens. Because the rooftop area that drained runoff to each rain garden was approximately five times larger than the area of the rain garden itself, evapotranspiration was a small percentage of the annual water budget. For example, during water year 2005, the maximum evapotranspiration of total influent volume ranged from 21 percent for the turf rain garden in clay to 25 percent for the turf rain garden in sand, and the minimum ranged from 12 percent for the prairie rain garden in clay to 19 percent for the prairie rain garden in sand. Little to no runoff left each rain garden as effluent and a small percentage of runoff returned to the atmosphere through evapotranspiration; therefore, the remainder was considered recharge. During water year 2005, recharge was 81 to 75 percent of total influent volume for the prairie- and turf-rain gardens in sand and 87 to 78 percent for the prairie- and turf-rain gardens in clay, respectively. Maximum recharge volumes ranged from 90 to 94 percent of the total influent volume in the turf and prairie rain gardens in sand and occurred during water year 2004. Maximum recharge in the turf and prairie rain gardens in clay ranged from 89 percent during water year 2007 to 98 percent during water year 2004. Median infiltration rates were an order of magnitude greater for rain gardens planted in sand than for those in clay, regardless of vegetation type. Under similar soil conditions, rain gardens planted with turf grass had lower median infiltration rates than those planted with prairie species. Median infiltration rates were 0.28 and 0.88 inches per hour in the turf and prairie rain gardens in clay, respectively, and 2.5 and 4.2 inches per hour in the turf and prairie rain gardens in sand, respectively. In general, infiltration rates were greater during spring (April and May) and summer (June through August) months. Of the six observed exceedences of the storage capacity of the turf rain garden in clay between April-November during 2004-07, five were predicted by use of a combination of the normalized surface storage volume, the median infiltration rate, and an estimate of specific yield for soils under the rain garden to a depth equal to the uppermost limiting layer. By use of the same criteria, in water year 2008, when the contributing drainage area to the prairie rain garden in clay was doubled, all four observed exceedences of the total storage capacity were predicted. The accuracy of the predictions of when the total storage capacity of the rain gardens would be exceeded indicates that by applying measurements of the appropriate soil properties to rain g