A study tested the hypothesis that the difference in mail returns to the 1990 census between related and unrelated households can largely be accounted for by differences in the way these households handle their mail. Data indicate that both household structure and mail handling behavior influenced census mail return.