In many modeling situations, a set of values for the model parameters is regarded as characterizing an individual. The modeler may, however, be interested in estimating the distribution of parameter values in the population from which the individuals are sampled. Some applications of population estimation, a survey of approaches to the estimation problem and some current work will be discussed. The approaches discussed include the "naive pooled data approach," two stage methods, the first order method proposed by Sheiner and Beal and a nonparametric maximum likelihood procedure. General approaches employing Bayes' Theorem are also considered.