The importance of testing for anticardiolipin antibodies (aCL) in the diagnosis of antiphospholipid syndrome (APS) in patients with thrombosis has recently been challenged (ISTH SSC meeting, Boston 2002). We have analyzed the antiphospholipid serology of 123 patients with persistent antiphospholipid antibodies (aPL) attending our hematology department. The cohort was tested for anti-β2-glycoprotein I (β2-GPI) antibodies and aCL of IgG and IgM class and for lupus anticoagulant (LA). Ninety-six of these patients fulfilled Sapporo clinical criteria for APS and 70 of these patients had venous and/or arterial thrombosis. Patients with LA plus anti-β2-GPI antibodies had significantly higher levels of IgG aCL and anti-β2-GPI antibodies than those exhibiting positivity for only LA or anti-β2-GPI antibodies (P < 0.05). Patients with aCL IgG levels over 60 GPLU were found in all cases to be positive for LA and anti-β2-GPI antibodies; 25.2% (31/123) of all patients and 26.04% (25/96) of patients fulfilling Sapporo clinical criteria for APS were positive for aCL only. The mean IgG aCL level in the Sapporo clinical criteria positive patients who had aCL only was 11.5 GPLU (normal < 5 GPLU). These data indicate that omission of aCL testing from the clinical investigation of APS could lead to a failure to diagnose the syndrome in a proportion of patients.