Evaluating the impacts of atmospheric dust on marine ecosystems and climate requires the use of three-dimensional transport models including a size-resolved bins scheme to describe the evolution of the particle size distribution. Recently, Forêt et al. (2006) proposed an alternative size bins scheme to better account for the physical size-dependent processes such as dry deposition. This paper evaluates the benefit of using this new bin scheme in three-dimensional transport models. This is achieved by performing a one-and-a-half month simulation with the CHIMERE-DUST model forced by the Mesoscale Model, Version 5 (MM5)/National Center for Environmental Prediction (NCEP) analysis fields, for an area covering the northern Atlantic Sea and the western Africa. Compared to a reference run performed using a large number of size bins (40), our results show that the size bins scheme proposed by Forêt et al. (2006) reduces at least by a factor of 2 the numerical errors on the simulated concentrations compared to the classical isolog bins scheme for the same number of size bins. However, with this new bins scheme requiring to define the bins according to a dry deposition velocity corresponding to a given friction velocity, we examine the errors associated to this constraint.