Dye tracing is a common and simple method for measuring surface velocity of overland flow. However, a correction factor is needed to convert the velocity of the dye front to mean velocity. Selecting the appropriate correction factor for particular conditions is complex and its behavior has not been well studied in sheet flow under rainfall on actively eroding surfaces. A series of simulated rainfalls were conducted on a 2 m by 6 m plot with gravelly sandy loam soil. The plot was progressively eroded to examine a wide range of hydraulic conditions. A total of 178 velocities (maximum and mean) were measured using electrolyte and dye tracers over three travel distances (1.65, 3.5, and 5.8 m), and three slope gradients (5%, 12% and 20%). Discharge, sediment yield, random roughness, rock cover, and a range of hydraulic variables were also determined. Mean flow velocity was derived from the centroid of the electrolyte breakthrough curve. Our findings suggest that the velocity correction factor is a dynamic, site specific property. A linear model was proposed to estimate the correction factor based on readily available predictor variables, those being travel distance, unit discharge, and maximum velocity. The results will benefit further investigations of overland flow hydraulics on semiarid rangelands by providing more comprehensive correction factors to determine mean velocity.