In preparation for routine anti-HTLV-III/LAV testing in the UK five commercial assays (A-E) were evaluated using 360 sera selected on clinical and epidemiological grounds. These comprised 220 specimens from blood donors, 83 specimens from patients in high-risk groups, and 57 specimens with features likely to produce false-positive results. Probably erroneous positive results arose from assay A in all three categories and assay B in the second and third categories. These reactions were much more common after specimens had been heated to 56 degrees C for 30 min. Except that an anti-HLA DR4,B5-containing serum was repeatedly positive by C, assays C, D, and E apparently did not give rise to false-positive results. Results by these three assays were also highly reproducible. In tests on serum dilutions the highest titres were obtained by assays A and D, but assays C and E discriminated most clearly between anti-HTLV-III/LAV positive and negative sera. These two assays were rapid and convenient and seemed particularly suitable for testing blood donations. Assay D was almost comparable with them in performance but more difficult to use. The commercial assays C, D and E, an antibody capture assay, and a simple immunofluorescence test could be the basis for a methodologically diverse national system of primary and confirmatory testing for anti-HTLV-III/LAV.