Lupus anticoagulant (LA) antibodies are acquired inhibitors of coagulation belonging-together with anticardiolipid (aCL) antibodies-to the family of antiphospholipid antibodies. Since LA antibodies affect coagulation reactions via recognition of the complex of lipid-bound prothrombin, they may be better named anti-prothrombin antibodies. We studied their immunological properties in the plasma of 59 patients with antiphospholipid antibodies by means of specific ELISA systems that allowed the characterization of the interaction of these antibodies with human prothrombin and anionic phospholipids. The mode of presentation of prothrombin was found to greatly influence the reactivity of anti-prothrombin antibodies. In fact, when plain polystyrene plates were used to immobilize prothrombin, virtually no binding was observed. Conversely, when prothrombin was coated on high-activated PVC ELISA plates, 34 samples (58%) contained antibodies that recognize human prothrombin in solid phase. In particular, IgG antibodies were found in 21 plasmas and IgM in 22; both IgG and IgM isotypes were present in 9 of these cases. A higher prevalence was observed in the ELISA for the detection of the antibodies directed at the calcium-mediated complex of phosphatidylserine (PS)-bound prothrombin: 53 samples (90%), preadsorbed with cardiolipin liposomes to remove aCL antibodies, showed the presence of IgG and/or IgM anti-prothrombin antibodies. When the results were analyzed according to the immunoglobulin isotypes, 44 (75%) and 39 (66%) samples were found to contain IgG and IgM anti-prothrombin antibodies, respectively. Both IgG and IgM were present in the plasma of 30 patients. Only half of these samples reacted also with PVC-bound prothrombin. Apparently, the higher rate of positivity of the ELISA for the detection of antibodies to the complex of PS-bound prothrombin was not due to differences in the amount of antigen available in the 2 systems, as judged by binding experiments performed with a rabbit polyclonal anti-human prothrombin antiserum. Finally, the anticoagulant properties of 14 total IgG preparations (12 of them contained anti-prothrombin antibodies positive in both ELISA systems, whereas the other 2 cases reacted either with PVC-bound prothrombin only or with PS-bound prothrombin only) were evaluated by diluted Russell's Viper Venom Time and by diluted activated Partial Thromboplastin Time. To rule out the beta 2-glycoprotein I (beta 2-GPI)-dependent anticoagulant effect of the aCL antibodies contained in the preparations, the coagulation tests were performed in beta 2-GPI deficient plasma. Six preparations failed to show anticoagulant activity in both assay systems, suggesting that 2 types of IgG anti-prothrombin antibodies exist, that differ with respect to their anticoagulant properties. These findings suggest that anti-prothrombin antibodies resemble aCL antibodies with respect to the behaviour in "in vitro" coagulation reactions and underline the wide heterogeneity of antiphospholipid antibodies.