Saliva specimens were tested for HIV antibody (anti-HIV) by an immunoglobulin G (IgG) antibody capture radioimmunoassay (GACRIA) and three sensitive commercial assays. In tests on 460 seronegative subjects and 196 seropositive subjects GACRIA was 99.8% specific and 100% sensitive. The Wellcome HIV monoclonal and Abbott recombinant DNA enzyme-linked immunosorbent assays (ELISAs) were also highly specific (99.8%, 100%) but they were less sensitive (90.9%, 82.0%). The Fujirebio particle agglutination assay was sensitive (97.8%) but its specificity was poor (84.1%). In testing saliva specimens from populations with an anti-HIV prevalence greater than 0.5%, sampling by GACRIA alone could provide a good estimate of the true prevalence. For true prevalences less than 0.5% good estimates could only be obtained if positive GACRIA reactions were confirmed by another independent salivary assay. Salivary testing for anti HIV is a convenient and potentially an accurate epidemiological tool.