This study evaluated accuracy of the assumption of the National Center for Educational Statistics (NCES) that higher education students of unknown race or ethnicity should be distributed over the remaining resident categories in the proportion of those whose race/ethnicity is known. This study endeavored to estimate the error associated with any inaccuracy in this assumption. Data were collected through a survey of 30 Missouri public two- and four-year institutions to determine institutional practices regarding unknowns before 1990 and through phone interviews of students with unknown race and ethnicity. Results from the university survey indicated that about 80 percent of institutions classified all unknowns as "white" suggesting that NCES's policy of proportionally distributing unknowns over the available resident categories did not accurately reflect the reporting practice among postsecondary institutions prior to 1990. Results from the phone interview survey of 102 students who failed to report their race/ethnicity found no African Americans in this group and more Native Americans, Asians, and Pacific Islanders than expected. The institutional and student surveys suggested that any assumption regarding the characteristics of unknowns has the potential to mislead policy decisions regarding current ethnic composition at institutions. Alternative methods of classifying unknown students are compared. (CK)