Runofi rates were estimated from rainfall rates and runofi amounts for 4 experimental sites in China, Malaysia, and Thailand before a physically based erosion model GUEST was used to determine the soil erodibility parameter and evaluate the potential to use the erosion model to predict the amount of soil loss on an event basis. We also examined 3 difierent ways of determining the soil erodibility parameter for the same storm event using: (i) hydrographs estimated from rainfall intensities and runofi amounts; (ii) an efiective runofi rate calculated from the hydrograph; (iii) an estimate of the efiective runofi rate based on a scaling technique involving the peak rainfall intensity and the gross runofi coe‐cient. All 3 methods can produce consistent soil erodibility parameters for a given runofi event. The calculated soil erodibility for individual storm events for all sites shows considerable temporal variation and for most sites a decreasing trend over time, as observed elsewhere in the same region. Among the 4 soils examined, the average soil erodibility tends to decrease as the ratio of coarse to flne materials decreases. When the erosion model GUEST is used to predict event soil loss using estimated soil erodibility parameters, an average model e‐ciency of 0¢ 68 is achieved for the sites tested.