BACKGROUND Human immunodeficiency virus (HIV) seroprevalence surveys help determine the extent of HIV and the need for routine HIV counseling and testing. We sought to describe trends in five south Georgia hospitals and compare two protocols for conducting anonymous HIV surveys. METHODS From 1993 to 1997, each hospital tested approximately 1,400 anonymous, leftover blood specimens per year. Two hospitals (C and E) tested all specimens, and three hospitals (A, B, and D) excluded blood known to be HIV-related. RESULTS The mean HIV seroprevalence at hospitals C and E was between 2.0% and 2.3% each year. The mean HIV seroprevalence at hospitals A, B, and D increased from 0.5% to 1.0% during 1993 to 1995, then decreased to 0.3% in 1997. CONCLUSIONS In hospitals C and E, the level of HIV disease was constant. In hospitals A, B, and D, the HIV seroprevalence decreased to 0.3%, suggesting that routine HIV counseling and testing is not currently needed in these hospitals. Hospitals should consider conducting periodic anonymous HIV seroprevalence surveys, sampling from all patients and distinguishing between specimens known to be HIV-related and those that are not.