The measurement of polydisperse populations of colloidal particles with quasielastic light scattering (QLS) is a common and dangerous practice. Here intensity autocorrelation functions with different linewidth distributions are simulated. The simulated data are treated with the histogram and exponential sampling methods and the results are contrasted with the results of the Laplace inversion routine of Provencher. All methods work well in calculating the size distribution from noise‐free correlation functions. In analyzing noisy data the methods may produce results that bear no relation to the true size distribution. The histogram method fails to determine size distributions from noisy simulated and real correlation functions. The exponential sampling method gives a qualitative measure of the size distribution when the required resolution does not exceed the limitations set by noise theory. Provencher's routine extracts smooth unimodal distributions very accurately even from noisy correlation functions, but determination of bimodal distributions may be unreliable. The importance of taking data at many scattering angles is emphasized and experimentally determined size distributions of samples of polystyrene latex spheres measured at multiple scattering angles are presented.