We discuss different ways to convert observed, apparent particle size distributions from 2D sections (thin sections, SEM maps on planar surfaces, etc.) into true 3D particle size distributions. We give a simple, flexible and practical method to do this, show which of these techniques gives the most faithful conversions, and provide (online) short computer codes to calculate both 2D-3D recoveries and simulations of 2D observations by random sectioning. The most important systematic bias of 2D sectioning, from the standpoint of most chondrite studies, is an overestimate of the abundance of the larger particles. We show that fairly good recoveries can be achieved from observed size distributions containing 100-300 individual measurements of apparent particle diameter.