DNA methylation patterns in blood samples are known to be influenced by a variety of exposures including lifestyle factors including smoking, alcohol consumption and diet. Although many of these exposures can be reasonably accurately assessed by self-report questionnaire, there are significant advantages to having objective measures of these exposures using biological samples over self-report which is known to be unreliable, especially for long-term exposures. We therefore propose to construct and validate an accurate DNA methylation-based model of long-term alcohol intake using data from a large number of cohort studies including ALSPAC.