Long-term monitoring is needed for direct assessment of soil organic carbon (SOC), soil, and nutrient loss by water erosion on a watershed scale. However, labor and capital requirements preclude implementation of such monitoring at many locations representing principal soils and ecoregions. These considerations warrant the development of diagnostic models to assess erosional SOC loss from more readily obtained data. The same factors aect transport of SOC and mineral soil fraction, suggesting that given the gain or loss of soil minerals, it may be possible to estimate the SOC ยฏux from the data on erosion and deposition. One possible approach to parameterization is the use of the revised universal soil loss equation (RUSLE) to predict soil loss and this multiplied by the per cent of SOC in the near-surface soil and an enrichment factor to obtain SOC loss. The data obtained from two watersheds in Ohio indicate that a power law relationship between soil loss and SOC loss may be more appropriate. When measured SOC loss from individual events over a 12-year period was plotted against measured soil loss the data were logarithmically linear (R 2 0 . 75) with a slope (or exponent in the power law) slightly less than would be expected for a RUSLE type model. The stable aggregate size distribution in runo from a plot scale may be used to estimate the fate of size pools of SOC by comparing size distributions in the runo plot scale and river watershed scales. Based upon this comparison, a minimum of 73 per cent of material from runo plots is deposited on the landscape and the most stable carbon pool is lost from watershed soils to aquatic ecosystems and atmospheric carbon dioxide. Implicit in these models is the supposition that water stable soil aggregates and primary particles can be viewed as a tracer for SOC.