Abstract Prostate-specific antigen (PSA) is the most used biomarker for prostate cancer detection. As well as non-specific, PSA may be non-sensitive - diabetes, obesity, and other components of the metabolic syndrome like hypercholesterolemia generally lower its levels, whereas hypertension increases them. Alcohol use is associated with these factors. Any lowering effect of alcohol on PSA, mediated through its effect on the above factors or otherwise, is thus of great public health importance, as it could delay prostate cancer diagnosis in a significant number of men. Alcohol may also cause prostate cancer, but in the era of PSA testing this could be masked by detection bias, failing to identify cases among heavier drinkers. The aim of this study is to investigate the relationship between alcohol intake and PSA levels in a sample of 10,000 white British men aged 50-70 years, without symptomatic prostate cancer. Data on PSA levels, alcohol intake and prevalent components of the metabolic syndrome were available from a random sample of 10,000 men, 1,800 of whom had been genotyped for a functional variant in the alcohol dehydrogenase gene ADH1B. We used linear regression to compare geometric means of PSA among groups with differing alcohol intake in the year before PSA testing, with and without adjustment for BMI and a history of diabetes, hypertension and hypercholesterolemia. We also compared PSA levels of carriers of the ADH1B*A allele, who have been shown to drink less, with those of non-carriers, to remove potential confounding often present in alcohol associations. After adjustment for age, BMI, centre, date of recruitment, and history of diabetes, hypertension and hypercholesterolemia, we found that PSA decreased with alcohol intake. Compared to men in the bottom third of the alcohol consumption distribution, those in the middle and top third had lower PSA (ratio of geometric means (RGM) and 95% confidence interval (CI): 0.96; 0.92-1.01 and 0.90; 0.85-0.94, respectively). ADH1B*A carriers had on average lower PSA levels than non-carriers (RGM 0.82, 95% CI 0.66-1.01). This genotypic effect seemed more marked for those consuming >= 10 units of alcohol a week, with geometric mean PSA among carriers 33% lower than among non-carriers (95% CI 5%-53%), whereas no effect was detected for light drinkers consuming <10 units/week (RGM 0.98, 95% CI 0.66-1.47) (p for interaction=0.07). Our analyses suggest the existence of a negative dose-response association of weekly alcohol consumption with PSA levels, independent of known determinants of PSA. Evidence of association with the ADH1B variant corroborates this finding and suggests a causal effect. This is the first large population-based study to show evidence that alcohol lowers PSA levels. While independent replication of the findings is warranted, sensitivity analyses are already in progress to investigate the effect of recent changes in alcohol behavior (“sick quitter” effect). Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 1839.