Accurate population estimates are critical to effective planning and policy. This study evaluates county population estimates using three different approaches to estimate domestic migration— typically the most volatile component of population change. Using the 2010 Census as a benchmark, it compares the standard net-migration approach to two gross migration approaches—a bi-regional approach and a multi-regional approach that models migration between county pairs. The bi-regional model produces the lowest average error and is a good choice when producing estimates for a large number of diverse areas. The multi-regional model works well for many counties, but is prone to extreme errors.