It was suggested that factor V Leiden and prothrombin G20210A mutations increase the risk of ischemic stroke only in combination with clinical risk factors of arterial ischemic disease. In these studies the controls were derived from the general population, with fewer clinical risk factors, which might have produced biased results. The factor V Leiden and prothrombin G20210A mutations were examined by polymerase chain reaction technique in 120 ischemic stroke patients and 120 controls younger than 65 years of age. Each patient had his own control, tightly matched in clinical risk factors. The prevalences of factor V Leiden and prothrombin G20210A mutations in patients were 8.3% (P = 0.02) and 7.5% (P = 0.04), respectively, and 2.5% for controls for both mutations. All carriers were single heterozygotes. In patients, but not in controls, the carriers of either mutation were mostly women and with fewer clinical risk factors for arterial ischemic events. In particular, considering both mutations as a single coagulation deficit, their presence increased the likelihood of ischemic stroke (odds ratio [OR] = 3.6; 95% confidence interval [CI] 1.4-9.3), especially among women (OR = 4.6; 95% CI: 1.2-17.8), normotensive persons (OR = 9.2; 95% CI: 1.1-17.8) and those having normal cholesterol (OR = 5.9; 95% CI: 1.6-21.2) and triglyceride serum concentrations (OR = 4.3; 95% CI: 1.5-12.8). In the studied sample of adult North Mediterranean population younger than 65 years the prevalences of factor V Leiden and prothrombin G20210A mutations were greater in patients with ischemic stroke than in matched controls. Unlike in studies with unmatched controls, we observed an apparently negative interaction of these mutations with clinical risk factors.