As part of an ongoing program of methodologic studies to estimate cancer incidence, prevalence, and survival at a small area level, we present the results of a hierarchical mixed effects model that predicts the number of cancer cases expected in each county, based on incidence data in 200 counties where the National Cancer Institute sponsors tumor (SEER) registries. We have modeled SEER age- and county-specific incidence data for tumor groupings as a function of age, corresponding mortality rates, and sociodemographic factors. Regression coefficients from these models were applied to non-SEER counties to predict incidence in these areas. These models were developed using a random subset of SEER county data, then validated by predicting incidence for the remaining SEER counties where observed data were available for comparison. The model for lung cancer among white males, for example, explains more than 95% of the variability in the data. Estimates of the expected number of incident cancer cases are useful measures of the cancer burden in the U.S. and provide information with which to monitor cancer trends and to plan cancer control activities. In addition, they may be used for quality control purposes, such as to assess completeness of reporting by individual non-SEER registries. Modifications to the initial models will be shown, including smoothed maps of new lifestyle covariates.