The size and shape of particles can be described using a 2D particle size distribution (PSD) where two characteristic lengths define each particle in the population. The determination of 2D PSDs based on microscopic pictures of particles in suspension is studied. The experimental data are represented as an axis length distribution (ALD) that can be extracted from a series of microscopic pictures by a fully automated image analysis. The problem of finding the underlying bi-dimensional PSD is stated as an optimization problem. For the solution a genetic algorithm is used. The approach is tested on simulated ALDs, as well as on an experimentally measured ALD obtained from carbon fiber particles. (c) 2008 Elsevier Ltd. All rights reserved.