When the atmospheric diffusion of a material is described by the usual Guysian plume model, dry deposition of the contaminant onto the underlying surface is commonly accounted for by appropriately reducing the source strength, as originally proposed by Chamberlain. A more realistic model is developed which selectively depletes the Gaussian plume in the vicinity of the deposition surface rather than throughout the vertical extent of the plume as done in the source depletion model. This improved model is used to show that the source depletion model consistently overpredicts the surface air concentration and the deposition at downwind locations close to the source and, as a consequence, is biased in the opposite sense for locations far from the source. At all distances from the source, the source depletion model overestimates the total deposition between source and receptor and consequently underpredicts the amount of remaining airborne material. Quantitative comparisons are shown to aid the user in choosing, for his particular circumstances, between the less accurate source depletion model and the computationally more complex surface depletion model. Dry deposition of an airborne material onto the un- derlying surface is of interest from two different stand- points. First of all, it can be an important sink for the material, reducing air concentrations (and conse- quent dry deposition) further downwind. In the case of a noxious substance this is beneficial to downwind receptors. However, dry deposition secondly is a mechanism for accumulation of the material on the ground and hence may be detrimental at the point of deposition. Since a high estimate for the deposition flux will be conservative at the point of deposition and, at the same time, nonconservative downwind of that point, modeling of the dry deposition process should be unbiased. However, it will be shown here that the source depletion model currently used for predicting deposition from the Gaussian plume is biased. This will be done by developing a surface dep- letion model which eliminates the artificial bias of the former model and by comparing the predictions of the two models. The biases of the source depletion model will be quantitatively delineated in order to provide a basis for choosing, in a specific situation, between the less accurate source depletion model and the computationally more complex surface depletion model.