An efficient and accurate method for the representation of particle size distributions in atmospheric models is proposed. The method can be applied, but is not necessarily restricted, to aerosol mass and number size distributions. A piecewise log-normal approximation of the number size distribution within sections of the particle size spectrum is used. Two of the free parameters of the log-normal approximation are obtained from the integrated number and mass concentration in each section. The remaining free parameter is prescribed. The method is efficient in a sense that only relatively few calculations are required for applications of the method in atmospheric models. <P style="line-height: 20px;"> Applications of the method in simulations of particle growth by condensation and simulations with a single column model for nucleation, condensation, gravitational settling, wet deposition, and mixing are described. The results are compared to results from simulations employing single- and double-moment bin methods that are frequently used in aerosol modelling. According to these comparisons, the accuracy of the method is noticeably higher than the accuracy of the other methods.