AIMS: To identify the threshold of alcohol consumption above which the balance of risk and benefit becomes adverse in diabetic subjects. METHODS: We studied demographic, lifestyle, dietary and clinical information in 216 hospitalized diabetic patients (171 men, 63 +/- 9 years old, 45 women, 67 +/- 5 years old) with a first event of an acute coronary syndrome (ACS) and 196 frequency matched (age-sex) diabetic controls, without any clinical evidence of coronary heart disease. Alcohol consumption was quantified and a measure for the comparisons was predetermined to be a wine glass (100 ml of wine, 12 g of ethanol) and its alcohol equivalents. RESULTS: Alcohol consumption was associated with an age-adjusted J-shape relationship with total cholesterol, blood pressure and smoking (all P < 0.001). A J-shape association was also found between alcohol intake and the risk of ACS (OR = 2.54-2.43 x (alcohol intake) + 0.80 x (alcohol intake)2, R2 = 0.96, P < 0.001), adjusted for several risk factors and interactions between alcohol intake and smoking status, job and familial stress, and low income. In particular, low alcohol consumption (< 12 g/day) was associated with a 47% (OR = 0.53, 95% CI 0.28-0.97) reduction of the prevalence of ACS, while a higher intake (12-24 and > 24 g/day) increased the prevalence by 2.7-fold (OR = 2.72, 95% CI 1.39-5.38) and 5.4-fold (OR = 5.44, 95% CI 1.21-24.55), respectively. CONCLUSIONS: Alcohol intake is a significant predictor of coronary events. Low-to-moderate intake seems to be associated with a reduction in the prevalence of ACS in diabetes, whereas higher consumption is associated with an increase in lipids and blood pressure levels, and also the risk of developing ACS.