A national survey to estimate prevalence of HIV infection would provide a useful addition to available information about the current status and future prospects of the AIDS epidemic in the United States. Both design and interpretation of a prevalence study should make use of existing information about the biology and epidemiology of AIDS. In conjunction with such information, a survey would help increase the accuracy of prevalence estimates and aid in making projections, despite the possibility that participation rates will differ in infected and uninfected individuals.