To compare the clinical results with the borreliacidal-antibody test (BAT) and two standard screening serologic tests for Lyme disease (LD)-the indirect immunofluorescence assay (IFA) and the enzyme-linked immunosorbent assay (ELISA).The medical records of patients from an endemic LD area, who had been serologically tested during the summer of 1992, were retrospectively categorized by clinical diagnoses without results of serologic tests. Serologic testing, which included control serum samples from patients from a nonendemic LD area, was performed in a blinded fashion, and the results were compared with the clinical categories.Medical records of 307 patients who had been serologically tested for LD were reviewed. We found untreated, active LD in 43 patients (early-localized LD, 21; early-disseminated LD, 14; and late-disseminated LD, 8) and treated LD in 33. Non-LD cases were categorized into acute or chronic conditions of unknown or known cause.Overall, the BAT had a sensitivity of 11% in active LD and did not correlate with results of other conventional surface antibody assays. The IFA and ELISA were more sensitive (67 to 93%), but false-positive results frequently were noted (20 to 40%) in acute and chronic non-LD inflammatory conditions. The specificity of the BAT, IFA, and ELISA in the control group was 96%, 93%, and 97%, respectively.Until the sensitivity, as measured by prospective clinical studies, is improved without loss of specificity, the BAT should not be used clinically for the diagnosis of LD. Suspected cases of LD with atypical clinical manifestations should have positive ELISA and IFA results confirmed with a standardized immunoblot assay.