Reported data about the effect of the 4G/5G PAI-1 polymorphism on plasma PAI-1 levels are controversial. This study was designed to determine the relative effect of the 4G/5G PAI-1 polymorphism on high plasma PAI-1 levels after adjustment for metabolic syndrome - related variables, and to test if this effect is modified by the smoking status. Six hundred and thirty one unrelated subjects (292 men; 35-74 years), from a cross-sectional population-based epidemiological survey in the province of Segovia (Spain) were studied. The higher frequency of high PAI-1 levels was found in 4G/4G subjects (5G/5G 19.4%, 4G/5G 21.6%, 4G/4G 33.7%, p = 0.003). A multiple regression model, adjusted for gender, age, BMI, waist circumference, triglycerides, HDL-cholesterol, HOMA IR and leptin, showed this adjOR: 4G/4G vs 5G/5G: 2.22, p = 0.008. When smoking status - 4G/5G PAI-1 interaction was included as an independent variable these results were not modified. Our results indicate that the 4G/4G PAI-1 genotype might be strongly associated with high PAI-1 levels regardless of metabolic syndrome-related variables and smoking status.