Public health surveillance is the cornerstone of public health practice. The uses of surveillance include the identification of patterns of health among population subgroups. The assessment of race and ethnicity in public health surveillance is fundamental to the reduction of preventable excesses in poor health among racial and ethnic populations. We review the use of race and ethnic variables in national public health surveillance systems in the United States. One barrier to the use of race and ethnicity in public health surveillance is the lack of scientific consensus on the nature of race and ethnicity and the measurement of these variables. Differences in terminology, data collection procedures, perceptions of group identity, and changing demographics of the U.S. population present particular challenges for surveillance. We propose criteria for any useful variables collected through surveillance. Application of these criteria to race and ethnicity suggests that race as assessed in surveillance is not primarily associated with biological characteristics, but it is more like ethnicity--a matter of self-perceived membership in population groups. Regular evaluation of surveillance systems will contribute to the usefulness of information on race and ethnicity in the improvement of the health of minority populations.