In the present study we examined the association between baseline alcohol consumption and 10-year mortality in subjects with normal and abnormal glucose levels (diabetes, impaired fasting glucose (IFG) or impaired glucose tolerance (IGT)). Furthermore, we assessed the 6-year cumulative incidence of diabetes in categories of alcohol consumption. In the Hoorn Study, which started in 1989, alcohol intake was assessed by questionnaire in 2393 subjects who were subsequently categorised into four groups of alcohol consumption (non-drinkers, up to 10 g per day, 10-30 and >or=30 g per day). Glucose tolerance status by oral glucose tolerance test was classified according to the WHO-1999 diagnostic criteria. Subjects who drank up to 10 g per day of alcohol had the lowest mortality risk. The age- and sex-adjusted mortality risks for non-drinkers were 1.55 (1.04-2.32) for subjects with normal glucose levels and 1.72 (1.05-2.82) for subjects with abnormal glucose levels. The risk of diabetes was also lowest for subjects who consumed up to 10 g per day: 8.0 versus 12.9% for non-drinkers (P<0.05). Higher alcohol intakes were associated with increasing risks for mortality and diabetes. Adjustment for classical cardiovascular risk factors and other lifestyle variables did not materially affect the estimates. In conclusion, moderate alcohol consumption was associated with a lower risk for mortality and diabetes.