Recent studies have examined the spatial heterogeneity of velocity fields over natural river boundaries. A key challenge is understanding how many velocity measurements are required to provide spatially representative estimates of the flow. This paper describes a series of laboratory experiments in which flow velocities have been measured in a detailed spatial pattern over two water-worked gravel beds. These data have been utilized to deduce the minimum density of measurements required to provide representative estimates of several spatially averaged flow parameters. This was coupled with an investigation into the influence of measurement density on the level of error in the estimation of these parameters. Empirical relationships are developed that can be used to estimate, a priori, the required minimum measurement density for a known precision and accuracy over macroscopically flat, water-worked gravel beds within a range of submergences. These estimates can be based on measurements of the bed roughness length scale, bed shear velocity, and flow depth alone. For all of the flow parameters, the level of error in adopting a lower measurement density than the required minimum, and its change with measurement density, was especially large at the lower densities. Given this dependence on measurement density, caution should be taken when comparing flow estimates from studies that have used similar flow and bed conditions but different measurement densities.