Based on a large US representative cohort with detailed baseline interview and examination data, the relationship between alcohol consumption and all-cause mortality is examined over a period of 15 years follow-up. Results show a significant linear relationship for females and males under 60 years of age at baseline, and a non-significant U-shape for the older ones. Both results remain stable for different kinds of adjustment including adjustment for nutritional variables and smoking. Excluding people with heart disease history at baseline leads to an even more pronounced linear relationship for both males and females under 60 years of age. Furthermore, it is shown that the curvilinear relationship for men found in previous research is partly due to the age groups examined.