Because HIV infection is not readily or easily ascertained, both survey methods and sentinel surveillance approaches must be used. At least some of the surveys must be anonymous and unlinked to identifiable persons to avoid the uninterpretable impact of a self-selection bias that could lead to significant underestimates and occasional overestimates of prevalence. Others must be unblinded, with careful interviews of voluntary participants to evaluate risk factors for HIV infection. Further, these surveys must continue over time to evaluate trends in infection. A comprehensive family of complementary HIV surveys and studies and a national household-based HIV seroprevalence survey have been undertaken by the Public Health Service in colaboration with other Federal agencies, State and local health departments, blood collection agencies, and medical research institutions. These projects focus on accessible segments of the general population, childbearing women, persons at high risk of HIV, and persons in special settings such as prisons and colleges. This surveillance approach will help monitor the levels and trends of HIV infection in the United States and held prioritize, target, and evaluate HIV prevention activities. 10 references. (Author abstract modified)