A simple model based on diffraction and geometrical optics relates the field-of-view dependent lidar signal returns from clouds to the size density distribution of their water droplets. The relation is written in a matrix form. The apparent size density distribution is obtained via matrix inversion with a constraint technique. The perturbation caused by the geometrical optics is interpreted in terms of small-particle diffraction. The apparent size density distribution is corrected for multiple scattering a posteriori. The correction is established by calculating the relative contribution of the different scattering orders to the broadening of the scattering phase function. The validity of the model is supported by Monte Carlo simulations and by experimental data performed under controlled conditions. Finally the inversion technique is applied to multiple-field-of-view (MFOV) lidar measurements from real clouds. © 1997 Society of Photo-Optical Instrumentation Engineers.