There are scant data available on the relationship between smoking and total prostate specific antigen, free prostate specific antigen and percent-free prostate specific antigen. Given the high prevalence of smoking and the frequency of prostate specific antigen screening, it is important to determine any association between smoking and prostate specific antigen values using nationally representative data. Included in the final study population were 3,820 men 40 years old or older who participated in the 2001-2006 NHANES (National Health and Nutrition Examination Survey) and met the eligibility criteria for prostate specific antigen testing. The distributions of total, free and percent free prostate specific antigen were estimated by sociodemographic and clinical characteristics. Multivariate linear regression models were fit to determine the adjusted relationship between smoking and total and percent free prostate specific antigen while simultaneously controlling for these characteristics. For all ages combined the median total and free prostate specific antigen levels were 0.90 (0.81-0.90) and 0.26 (0.25-0.28) ng/ml, respectively. Multivariate linear regression analysis showed that total prostate specific antigen was 7.9% and 12.2% lower among current and former smokers, respectively, than among never smokers. High body mass index and diabetes were also statistically significantly associated with a lower total prostate specific antigen. Approximately a third of the men had a percent free prostate specific antigen less than 25%. Current smokers had a significantly lower percent free prostate specific antigen than former smokers. Our finding that smoking is inversely associated with total prostate specific antigen may have potential implications for the interpretation of prostate specific antigen levels in men who are current or former smokers. Given the high prevalence of smoking, obesity and diabetes, additional research on the combined effect of these health risk factors is warranted.