HLA-B27 typing contributes to the diagnosis of ankylosing spondylitis. The classical technique of microlymphocytotoxicity is costly and can give false-negative results. We have compared 304 samples using two relatively new methods - flow cytometry and PCR-SSP - and evaluated their respective uses in routine analysis. Flow cytometric HLA-B27 testing was performed using three monoclonal anti-B27 antibodies (HLA-ABC-m3, GS145.2 and FD705 clones). Cut-off values were established to differentiate HLA-B27-positive from HLA-B27-negative samples with ROC curves. Although flow cytometric analysis with a reliable monoclonal antibody (mAb) is valuable for HLA screening, none of the HLA-B27 flow cytometry protocols was sufficient on its own to ascertain the HLA phenotype in 100% of samples. Two false negatives were observed with the FD705 mAb and the use of two different monoclonal antibodies did not increase the accuracy of HLA-B27 typing. HLA-B27 typing using molecular biology is a reliable but costly technique. Therefore we suggest that DNA typing could be used as a complementary technique and applied to samples whose HLA-B27 phenotype cannot be determined by flow cytometry. The association of flow cytometry and DNA typing is, in our experience, an economical and reliable approach.