Laboratory flume experiments were carried out to evaluate the effect of particle density on bedload transport of sand-sized particles and the effect of a suspended load of clay particles (kaolinite) on bedload transport of sand-sized particles in rill flow conditions. Three materials in the range 400–600 μm were selected to simulate bedload transport of primary particles and aggregates: sand (2650 kg/m3), crushed brick (2450 kg/m3) and anthracite (1300–1700 kg/m3). In the two first experiments, two different methods were applied to determine bedload transport capacity of coarse particles for various conditions of flow discharge (from 2 to 15 L/min) and slope (2.2, 3 and 4%). In the third experiment, clear water was replaced with kaolinite–water mixture and bedload transport capacity of crushed brick particles was determined for a 4% slope and different concentrations of kaolinite (0, 7, 41 and 84 g/L). The results showed that bedload transport increased significantly with the decrease in particle density but the effect of particle density on transport rates was much less important than flow discharge. Velocity measurements of clear flow, flow mixed with coarse particles and coarse particles confirmed the existence of a differentiation between suspended load and bedload. In these experimental conditions, suspended load of kaolinite did not affect bedload rates of crushed brick particles. Three transport capacity formulae were tested against observed bedload rates. A calibration of the Foster formula revealed that the shear stress exponent should be greater than 1.5. The Low and the Govers unit stream power (USP) equations were then evaluated. The Low equation was preferred for the prediction of bedload rates of primary particles but it was not recommended in the case of aggregates of low density because of the limited experimental conditions applied to derive this equation. Copyright © 2008 John Wiley & Sons, Ltd.