Prothrombin is a key factor in blood clotting, a process intimately involved in thrombotic disease. We assessed prothrombin levels and G20210A genotype in a case-control study within the Cardiovascular Health Study. Cases included angina, myocardial infarction, stroke, and the presence of MRI-detectable infarcts (n approximately 250 each). Population-based controls free of clinical cardiovascular disease (CVD) (n approximately 500) and a subset free of clinical and subclinical CVD (n approximately 250) were used for comparison. The 20210 A allele, frequency 2.9%, was associated with higher mean prothrombin levels: 166.3 vs. 139.5 microg/ml (P <0.001). Significant correlates of prothrombin included gender, plasma lipids, other vitamin K-dependent proteins, and inflammatory markers, but not race, smoking, hypertension, diabetes, measures of subclinical CVD, or markers of procoagulant activity. Compared to controls, neither genotype nor prothrombin level was associated with any CVD case group. We conclude that, in the elderly, neither prothrombin level nor 20210 genotype were associated with either CVD risk factors or events. This is consistent with the lack of association of prothrombin levels with measures of underlying CVD or procoagulant markers.