Abstract This paper describes work undertaken to demonstrate the applicability of artificial neural network technology to the classification of aerosol particles in terms of shape and size. The neural network demonstrator software has been developed by EDS-Scicon, with experimental data and expert domain knowledge provided by the Chemical and Biological Defence Establishment. Data on a range of aerosol particles with well defined shapes, and characteristic sizes, has been generated using a prototype light scattering instrument. This instrument monitors the cylindrical symmetry of laser light scattered by single aerosol particles using three photomultipliers placed around the optical axis. It also uses the narrow forward scattered light to obtain a measure of the size of the particles. Graphical representation of the data enables classification of the particles by visual inspection of the plots. The pattern classification capabilities of neural networks make them well suited to the shape classification of aerosol particles based on this light scattering data. This paper describes work undertaken to demonstrate the ability of neural networks to perform the classification task automatically when trained on samples of the photomultiplier output. The background to the application is described, along with details of the network architecture and the performance achieved by the fully trained network.