Dispersal is a critical process in plant populations determining the genetic structure within populations, the connectivity among populations, and the demographic ability to colonize new site. Numerous methods have been proposed for modeling the dispersal curve, but the choice of method for estimating patterns of dispersal remains one of the most enigmatic problems in plant ecology.  For pollen movement specifically, several forms of dispersal curves can be fit to data for inter-parent distances, but it is rare that any one of these models is clearly superior for a given data set.  The major methodological issues confounding this approach are large variation in the size of the tail expressing longer-distance matings and the paucity of data at these distances.  Here we propose a novel method that may help avoid this so-called "fat tail" problem, by focusing upon the estimation of among-individual variation in pollen dispersal rather than within-population mean pollen dispersal.  We model pollen dispersal via a pseudo-Bayesian hierarchical approach that allows for great flexibility in expressing variation, and estimate parameters using Markov Chain Monte Carlo.   We test this method using simulated data and an existing dataset derived from progeny of a wind-pollinated tree.  We also provide software for estimating pollen movement using R/S-Plus.  Among other benefits, results gained from this approach are particularly well-suited to the construction of realistic population simulations, with realistically structured dispersal variation within and among populations.