For wind erosion studies the distribution of terminal velocity (Wt) is a far more relevant soil property than the particle size distribution. An air settling tube was constructed to measure the distribution of Wt for the soil particles in a given sample. The apparatus consists of a vertical tube (height h=6.09 m) down which a small soil sample is dropped onto a weighing device which records the accumulated mass as a function of time t since release of the sample; this yields the distribution of average fall velocity (Wa=h/t) for the sample. A maximum sample size to avoid clustering effects was determined empirically. To convert Wa to Wt, a correction for initial acceleration is needed: three correction methods, at different levels of simplification, were developed. The settling tube was then tested with four different kinds of material. (1) Glass spheres were used to verify the acceleration correction methods experimentally: the most sophisticated method worked well for all sphere diameters and the simplest for diameters less than 0.25 mm. (2) Sand samples, sieved into size classes with median sieve diameter ds, had Wt values equal to those of spheres with equivalent sphere diameter d=0.9 ds, with a scatter of ¿20% dependent on dryness. (3) For organic material, Wt could be approximated (to ¿30%) by its value for an equivalent sphere diameter d equal to the cylinder diameter for cylindrical or fibrous particle forms, the particle length for cubelike forms, and 0.5 of the square root of area for platelike forms (4) For field soils, Wt distributions depended on wind erodibility: the distributions were unimodal and sharply peaked near 1 m s-1, while for nonerodible soils they were broad and tended to a bimodal form with the lower peak near 0.3 m s-1. For the aggregated nonerodible soils, size distributions inferred from the settling tube differed substantially from sieved size distributions, yielding more large particles and fewer small particles. ¿ American Geophysical Union 1991