Serological typing with the microlymphocytotoxicity test (MLCT) and flow cytometry (FC) using HLA-B27 antisera is commonly used for the determination of HLA-B27. However, in some patients tested more than once, negative results have turned out to be positive at following investigations. We retested by polymerase chain reaction (PCR) samples from 20 randomly selected patients with reactive arthritis or Reiter's syndrome who had now been followed for 20 yr. Ten of the patients were originally tested to be HLA-B27 positive and 10 HLA-B27 negative by the MLCT. All 10 serologically HLA-B27 positive individuals were also positive in the PCR. However, 2/10 patients interpreted as being HLA-B27 negative were positive by PCR. At this time, the same two patients were also positive in the routine MLCT and FC using four different monoclonal antibodies against HLA-B27. PCR is superior to serological techniques to determine HLA-B27 positivity unequivocally, since it is based on the detection of HLA-B27 gene sequences.