At finer scales, raindrops are the sources of the onset of soil erosion. Understanding the effects of raindrops at the decimeter scale is useful for soil erosion prediction, understanding erosion principles, and deriving erosion control management practices. The objective of this study was to develop and rest a physically based model to predict the effect of raindrop erosion on soil microtopography and identify the parameters that can be experimentally measured. The model has three parameters: (i) detachment rate mu similar to (9.0 +/- 4.0) x 10(-2) kg m(-2) mm(-1), (ii) average projection distance lambda similar to 0.15 +/- 0.05 m, and (iii) a dimensionless anisotropy coefficient delta similar to 3 +/- 1, which expresses the slope dependency of lambda and mu. Variation in soil height caused by raindrop erosion followed a diffusion-type equation with a source term. Under uniform conditions of soil and rainfall, the model simplifies into a basic diffusion equation. Under the homogeneous bare soil condition, soil surface roughness is predicted by an exponential decay model. Under nonuniform conditions, when sparse perennial vegetation protects the soil locally from raindrop impact (a common surface feature in semiarid areas), the model predicts that small mounds of 2 to 30 cm in height can develop underneath the cover. On a horizontal surface, the mound height asymptotically tends to a limit. On sloping areas, however, mounds are predicted to develop faster, higher, and to be asymmetric. Under both flat and sloping terrain, model predictions were found consistent with published data and models, with the noticeable improvement that the model parameters can be measured by laboratory experiments.