Measuring the size distribution of dust particles is of interest in many scientific and technological contexts. One of the most widely used techniques is laser light scattering (LLS), which provides the distribution of surface-equivalent spheres that fits the observed angular dependence of light scattered by a sample. We have revisited the problem of the uncertain lower size limit of this method by simulating laboratory measurements of the light intensity scattered by polydisperse spheres and irregular particles (agglomerated debris and pocked spheres), from which the original distributions are retrieved by regularized inversion with Mie and Fraunhofer phase functions. For the usual combination of blue (λ = 466 nm) and red (λ = 633 nm) light sources, size distributions of spheres with radii r > 0.1 μm are retrieved with Mie if the true complex refractive index (m = n – ik) is used. The retrieval for 0.1 μm < r < 3 μm is sensitive to errors in the assumed m, which results primarily from the dependence of the scattering efficiency Qsca on m. Irregular particle shape has also an impact on the Qsca vs. r curves, whose maxima are shifted towards larger r and are smoother compared to spherical particles. For a violet-blue wavelength (λ = 442 nm), good retrievals are obtained for irregular particles with r > 1 μm even if m is not very well known or the Fraunhofer model is used. Spurious slumps and enhancements appear for r < 1 μm, although if n is known, the actual lower limit decreases for increasing n. This implies that LLS size distributions of submicron irregular particles may not be accurate. Establishing the lower size limit requires inspection of the dependence on m and analysis of the irregularity of samples.