Densiometric quantitation and spreadsheet normalization were used to refine the parameters defining a treponemal Western blot (TWB) test for syphilis. Initially using 84 defined reactive and 105 defined non-reactive sera, we determined that the immune response to the 17 kDa antigen was the most critical of the following three candidate test determinants: the 47 kDa, 17 kDa and 15.5 kDa bands. In a second study using 124 cases of clinically diagnosed syphilis and 354 "normal" donors, a diluted serum sample was included as a minimal reactive control for the 17-kDa immune response. Reactivity to all three test determinants was obligatory for a test result to be interpreted as positive. Of the 124 cases of syphilis, 7 were nonreactive by TWB (sensitivity = 94%); of the 354 normal donors, 7 tested reactive (specificity = 98%). Forty (11%) normal serum samples had detectable but less than minimal reactivity to the 17 kDa band. Frequencies of immune response to a larger group of 12 antigens were tallied for the 124 clinically diagnosed cases of syphilis and an equal subset (124) of the normal group. In the normal subset, 72% and 52% of the samples had detectable reactivity to the 47 and 15.5 kDa antigens, respectively, while 10%, 5% and 3% reacted with the 17, 24 and 44.5 kDa antigens, respectively. Follow-up TWB testing of the clinically diagnosed cases revealed that previously untreated patients with primary or secondary syphilis were more likely to a show decrease in TWB reactivity than patients with latent symptoms who had been treated previously. As a diagnostic indicator of syphilis, the 17-kDa antigen was found to have the best combined attributes of sensitivity and specificity. Although, the highly specific 44.5 kDal and 24 kDal bands were often redundant as diagnostic indicators they are useful for the interpretation of borderline results. In addition, absence of the highly sensitive 47 and 15.5 kDa indicators should be useful in resolving some problem diagnoses.