The statistical properties of potential estimators of forest area for the USDA Forest Service's Forest Inventory and Analysis (FIA) program are presented and discussed. The current FIA area estimator is compared and contrasted with a weighted mean estimator and an estimator based on the Polya posterior, in the presence of nonresponse. Estimator optimality is evaluated both theoretically and via simulation under bias and mean squared error criteria.The results indicate that, under realistic conditions, the current FIA area estimator can sometimes result in substantial bias and have a higher mean squared error than both of the alternative estimators. This finding is of special interest because the same factor that contributes to this increased bias and variance applies to all area-based FIA estimates. The weighted mean and Polyha posterior estimators gave similar results for estimating the total area of a domain. It is concluded that the main advantage of the latter approach is that many other statistics are obtainable because the entire population distribution is estimated from the same sampling effort. The cost of this advantage for the Polya posterior approach is that a single result requires many more computer operations, a cost that has become virtually ignorable over the past decade.