Three soils were exposed to three consecutive simulated rains under dry, wet, and water table conditions with gypsum as a tracer placed at a 5-mm depth to evaluate the extent and nature of chemical transfer from soil solution to runoff and to determine the effective depth of the mixing zone. No electrolyte release was detected in runoff during the dry run for any of the soils. Electrical conductivity decreased exponentially with time during the initial stages of the wet and water table runs. Results were consistent with the complete mixing concept, but they also suggest that the assumption of no chemical transfer into the mixing zone from below should be modified. The effective mixing depth appears to be less than 3--4 mm. Two timescale processes were identified. The fast rate process, driven by raindrop impact and confined to the mixing zone, causes an exponential depletion of chemicals from that zone. The slow rate process, dominated by molecular diffusion and mechanical dispersion, describes chemical transfer to the mixing zone from below. The identification of the two processes indicates that the fast rate process is adequate for approximating chemical loss under free drainage conditions, while otherwise the slow rate process must be considered.¿ 1997 American Geophysical Union