This one is a good read: an old Harvard Medical School article about what they term "psychosocial benefits of moderate alcohol consumption".

http://www.peele.net/lib/benefits.php (by the way the website is about addictions)

They identified, back in the 1990s, a lack of consensus in what constitutes moderate drinking for optimal health and longevity.

"What Constitutes Moderate Drinking?
The definition of optimal drinking for health purposes has been fairly well accepted in the United States and United Kingdom as one or two drinks daily and at the lower end of the range for women (Department of Health and Social Security, 1995; U.S. Departments of Agriculture/Department of Health and Human Services, 1995). Going farther afield, however, this range may expand. Grønbæk et al. (1995) found mortality-rate gains for men and women up to three to five drinks of wine daily in Denmark, as did Fuchs et al. (1995) with a sample of women in the United States. Doll (1997) summarized various factors contributing to the relativity of both reported and optimal unit consumption across national boundaries, including extremely different definitions of what constitutes a standard drinking unit and the common underreporting of alcohol consumption. Thus, in Poikolainen's (1995) cross-cultural review, minimum mortality was associated with consumption levels ranging from one to five drinks daily. In the current review, which includes ethnographic as well as epidemiologic data and covers a range of cultural settings beyond those involved in typical epidemiologic studies, moderation is an even more variable ideal."

I see that even now in 2013 moderation is still a variable idea. There still is no consensus.