Experiments were undertaken to determine the feasibility of tracing sediment movement in interrill overland flow. Crushed magnetite was introduced as a source-line 10 cm wide by 8 m long on a runoff plot 18 m wide by 29 m long located in southern Arizona. Initial magnetic susceptibilities along this source line, and along three transects located 0·25, 2·95 and 5 m downslope of the source-line, were measured. Movement of the magnetite in response to three rainfall simulation experiments was monitored. During the first two experiments, overland flow discharge was sampled at miniature flumes located along two cross sections on the plot downslope of the source-line, and at a supercritical flume at the plot outlet. Magnetic susceptibilities along the source-line and transects were measured after all three experiments. Results show that the magnetite moves very early in the experiments and that it reaches one of the flumes 2 m downslope of the source-line in 3 min. Most of the tracer moves a very short distance: 29·7 per cent is deposited within 25 cm of the source-line and only 2·2 per cent is deposited 2·95 m away. The deposition rate appears to decrease exponentially away from the source-line. Very little magnetite is recorded in the flow through the miniature flumes: in general it makes up less than 1 per cent of the total sediment load. No temporal pattern in these percentages is observed. Magnetite appears to be an effective tracer of sediment movement in interrill overland flow, though its higher density than natural soil may affect its detachment and transport.