Most pF curves are shaped as cumulative distribution functions. From this observation, restricted to nonshrinking or swelling soils, a close relationship is established between the moisture characteristic and the pore size distribution, based on an empirical law relating the pore suction to a characteristic size referred to as the effective pore radius. For many soils showing a moisture characteristic similar to a normal cumulative distribution function, the derived pore size distribution is log normal. In this case, a direct identification technique is developed, yielding the parameters of the density function from the experimental pF-? relationship. Practical applications demonstrate the validity of this probabilistic model, which gives good agreement with morphometric pore size data, available in the literature. The model also gives rise to an analytical expression for the pF curve in terms of two structural parameters of the porous system, the mean and the standard deviation of the effective pore radius.