Rills often act as sediment sources and the dominant sediment and water transport mechanism for hillslopes. Six experiments were conducted on two soils and a uniform sand using three experimental methodologies. The results of this study challenge the assumption often used in hydrologic and erosion models that relationships derived for sheet flow or larger channel flow are applicable to actively eroding rills. Velocity did not vary with slope, and Reynolds number was not a consistent predictor of hydraulic friction. This result was due to interactions of slope gradient, flow rate, erosion, and the formation of rill roughness, bed structures, and head cuts. A relationship for rill flow velocities was proposed. Stream power was found to be a consistent and appropriate predictor for unit sediment load for the entire data set, while other hydraulic variables were not. The data for stream power and sediment load fit the form of a logistic curve (r2 = 0.93), which is promising relative to recently proposed erosion models which are based on probabilistic particle threshold theory.