AIM: To determine if a scheme for validating enzyme immunoassay (EIA) results could be devised that did not require costly and methodically elaborate supplemental assays. METHODS: Samples (n = 525) from patients with haemophilia A, leukaemia, and chronic liver disease and at increased risk of hepatitis C virus infection were tested by EIA-1 (Ortho Diagnostics), an assay which uses recombinant HCV fusion proteins as antigens, and by EIA-2 (United Biomedical), an assay based on synthetic HCV oligopeptide antigens. RESULTS: Samples (n = 193) were repeatedly reactive in both EIAs. Of these, 190 (98%) yielded reactivities in both of two supplemental assays used, one an immunoblot assay (RIBA) using recombinant HCV polypeptides similar to EIA-1 antigens, and the other a neutralisation EIA (EIA-2N) based on antigenic competition with HCV peptides similar to EIA-2 antigens. The three samples not reactive in supplemental tests exhibited low EIA optical density (OD) values (signal/cutoff ratios of less than 3). Hence, all specimens reactive and yielding high OD values in both EIAs were also reactive in supplemental assays. Twenty four samples were reactive in EIA-1 only and nine (38%) of these were reactive in RIBA. Fourteen of the 15 (93%) specimens reactive in EIA-1 but not RIBA were derived from patients with chronic liver dysfunction. Two samples were reactive in EIA-2 only, of which one was reactive in EIA-2N and none in RIBA. CONCLUSIONS: Compared with EIA-2, EIA-1 yielded more validated reactive samples and resulted in more non-validated reactivities. It is therefore suggested that for clinical diagnosis: (i) EIA-1 be used for anti-HCV testing and EIA-2 for validation of EIA-1 reactivities; (ii) samples concordantly reactive in EIA-1 and EIA-2 and displaying high OD readings be considered HCV antibody positive without supplemental testing; (iii) supplemental testing by RIBA be limited to samples reactive in EIA-1 but equivocal or unreactive in EIA-2 and those concordantly reactive but exhibiting low absorbance readings.