Rapid, on-site human immunodeficiency virus (HIV) testing has the potential to improve the delivery of prevention services in publicly funded counseling and testing sites. The Single Use Diagnostic System (SUDS) HIV-1 is the only rapid enzyme immunoassay (EIA) approved for diagnostic use in the United States. To evaluate the feasibility of using SUDS in public clinics and to validate the test's performance in a public health laboratory, we conducted blinded SUDS testing on plasma sent for HIV testing. From 19 March through 30 June 1993, 1,923 consecutive samples from a sexually transmitted diseases clinic and an HIV counseling and testing clinic were tested on site with SUDS. Tests done in the first two weeks with a malfunctioning centrifuge n = 402) and those done when there were excessively high temperatures in the laboratory (n = 53) were analyzed separately. Of 1,466 tests, 39 were positive by both SUDS and EIA (with Western blot [immunoblot] confirmation) and 7 were SUDS positive and EIA negative. Western blotting was used as the "gold standard" to adjudicate these discrepancies. There were no SUDS-negative and EIA-positive tests. Compared with that of EIA (with Western blot confirmation), the sensitivity of SUDS was 100% (95% confidence interval, 88.8 to 100%) and the specificity was 99.5% (95% confidence interval, 98.9 to 99.8%). The positive predictive value of SUDS was 88% in the STD clinic and 81% in the HIV counseling and testing clinic. There was a 7.7-fold increase in false positives, from 0.48 to 3.7%, when there was inadequate centrifugation and when the temperature exceeded the manufacturer's recommendations. Rapid, on-site HIV testing by the SUDS assay is feasible and practical in public health settings. The test can be performed accurately, at reasonable cost, and within the time frame of a typical clinic visit. Caution should be used, however, as two conditions adversely affected the accuracy of this test: inadequate specimen preparation and elevated temperature.