Anti-centromere antibody (ACA) has been reported to be associated with Sjogren's syndrome (SS) and the clinical significance of anti-CENP-H antibody remains unknown. To determine the clinical significance of anti-CENP-H and anti-centromere antibodies in primary SS, sera from 62 patients with primary SS and 40 normal controls were examined for anti-SS-A/SS-B antibodies, ACA and anti-CENP-H antibodies, by enzyme-linked immunosorbent assay and indirect immunofluorescence (IIF), respectively. Of the 62 serum samples with primary SS, 17 were positive with ACA and anti-CENP-H antibodies. Sera from SS patients with anti-CENP-H and ACA antibodies do not contain anti-SS-A/Ro and/or anti-SS-B/La antibodies. No anti-CENP-H antibody was found in sera of normal controls. An increased frequency of ACA and anti-CENP-H antibodies was found for the first time in patients with SS. Anti-CENP-H antibodies and anti-SS-A/Ro or anti-SS-B/La antibodies are present mutually exclusive. Patients with anti-CENP-H antibodies had a lower frequency of rheumatoid factor (RF). SS can be subdivided serologically into two groups; group one with anti-SS-A/Ro and/or anti-SS-B/La antibody, group two with ACA and/or anti-CENP-H antibodies. We recommend that ACA or anti-CENP-H antibodies should be considered as one of the serological markers for SS.