To clarify the relationship between alcohol consumption and type 2 diabetes we conducted a meta-analysis of published epidemiological studies. Data from 13 cohorts were included in the analysis. The results of these studies are consistent with regard to moderate alcohol consumption, indicating a protective effect in the order of 30% (relative risk [RR]meta = 0.72, 95% CI = 0.67-0.77). The reduced risk is seen in men as well as in women, although few studies investigated women. No protective effect of high alcohol consumption was seen and one cannot rule out that large intakes of alcohol may increase the risk of type 2 diabetes. Results from published studies suggest a U-shaped relationship between alcohol and type 2 diabetes, but this is based on rather few studies with heterogeneous design and definitions. It seems important to further investigate if, and to what extent, high alcohol consumption increases the risk of type 2 diabetes. Aspects of moderate alcohol consumption also need further investigation; these include type of drink, frequency of drinking, sex and ethnic differences.