To determine whether saliva could serve as an alternative to serum for HIV-antibody testing in an ongoing sentinel surveillance program in Thailand. Serum and saliva specimens were collected from 1955 individuals in four of the 73 sentinel sites of the national surveillance program in Thailand. Intravenous drug users, female prostitutes, and men attending sexually transmitted disease clinics were included as participants. All specimens were collected and tested anonymously. Saliva was gathered with the Omni-Sal collection device and analyzed for the presence of HIV antibodies using the immunoglobulin G antibody-capture enzyme-linked immunosorbent assay (GACELISA) laboratory test, specially designed for low concentration body fluids. Our gold standard was serum, collected and analyzed independently from the saliva specimens, using an ELISA test for screening and Western blot for confirmation. Linkage between serum and saliva was blind to the laboratory. A set of HIV-positive and HIV-negative quality assurance samples for both serum and saliva were also analyzed blind. Findings are presented as observed in the field, and as quality assurance samples after the correction of various field and laboratory errors. The sensitivity of the GACELISA with saliva was 98.0% in the field (298 HIV-positive specimens), 100% after correction of errors (300 HIV-positive specimens), and 100% among the quality assurance samples (95 HIV-positive specimens). The specificity of the GACELISA was 99.4% in the field (1653 HIV-negative specimens), 99.6% after correction of errors (1654 HIV-negative specimens), and 100% among the quality assurance samples (96 HIV-negative specimens). Our findings support other published studies that also featured the GACELISA. We conclude that saliva is comparable to serum for assessing HIV antibodies in individuals for surveillance and screening purposes.