In October 1994, the Second National Conference on the Serologic Diagnosis of Lyme Disease recommended a two-step approach to serological testing. The first step was the performance of an enzyme-linked immunosorbent assay (ELISA); the second step was a confirmatory immunoblot. New criteria for the interpretation of a positive immunoblot were also recommended. The committee decided to omit the 31- and 34-kDa bands (OspA and OspB, respectively) from the choice of bands considered diagnostic for a positive immunoblot. Since we had previously included these in our diagnostic criteria for Lyme disease-positive immunoblots, we reviewed data for all patients attending a Lyme disease center with positive ELISAs and immunoblot assays for Lyme disease from 1 September 1992 to 31 December 1993. The criteria for a positive Western blot (immunoblot) were the presence of 5 or 12 bands, including the 10 recommended by the conference, and the presence of the 31- and 34-kDa protein bands. Of the 136 patients evaluated, 50 were considered to have Lyme disease. Of these 50, 4 (8%) would not have met immunoblot criteria for the diagnosis if the new recommendations were used. Had the 31- and 34-kDa bands been included as part of the diagnostic requirements for immunoblot, these patients would have been included. Although overdiagnosis of Lyme disease appears to be the more frequent problem, our concern is that the exclusion of the 31- and 34-kDa protein bands from the diagnostic criteria may result in the underdiagnosis of Lyme disease by those who would rely too heavily on serological confirmation. The addition of the 31- and 34-kDa bands to those recommended for confirmatory immunoblot should be reconsidered.