Antiphospholipid antibodies in patients with "antiphospholipid syndrome" may be directed at least in part against plasma phospholipid-binding proteins, such as beta 2-glycoprotein I or prothrombin, which are involved in the control of thrombosis and haemostasis. IgG-class antibodies against prothrombin and beta 2-glycoprotein I were measured by enzyme-linked immunoassay in initially healthy middle-aged dyslipidaemic men (non-high-density lipoprotein > 5.2 mml/l). Serum samples had been drawn at entry to a 5-year coronary primary-prevention trial with gemfibrozil from 106 subjects who experienced either a non-fatal myocardial infarction or cardiac death during the follow-up and from 106 subjects without coronary episodes, matched for treatment group (gemfibrozil/placebo) and geographical area. The antiprothrombin antibody level, as expressed in optical density units, was significantly higher in patients than in controls (0.26 +/- 0.17 versus 0.22 +/- 0.09; p < 0.02). A high level of antiprothrombin antibodies (highest tertile of distribution) predicted a 2.5-fold increase in the risk (95% confidence interval 1.2-5.3) of myocardial infarction or cardiac death. The distribution of IgG-class antibodies against beta 2-glycoprotein I did not differ significantly between cases and controls. The joint effect of antiprothrombin antibodies and other factors associated with hypercoagulative state: triglyceride level, lipoprotein(a) and smoking, was multiplicative for the risk. Antiprothrombin antibodies are a new immunological predictor of myocardial infarction and the effect of these antibodies may be mediated by hypercoagulative mechanisms.