OBJECTIVES: A recent study showed that 28% of individuals in France are occasional heavy drinkers with short term social consequences that are underestimated. The objective of this study was to construct several risk scores to predict the probability of having drinking related social consequences. METHODS: Data from wave 1 and 2 of National Epidemiologic Survey on Alcohol, separated into a construction and validation samples, were used to predict the risk of dependence and alcohol-related depression, legal, relational and work issues in the next 3 years. Patients were included in the analysis if they declared drinking alcohol at least once in the last 12 months. Alcohol consumption was recoded in three variables corresponding to the AUDIT-C questions: drinking frequency in the last year, average drinks consumed when drinking and frequency of consuming five drinks or more. The probability of social consequences were modeled in the construction sample using logistic regression, with age, sex and alcohol consumption variables as predictors. Goodness of fit was assessed in the validation sample based on C-index and R².  RESULTS: Based on only five predictors, the risk equation demonstrated a good goodness-of-fit with a c-index of 0.87 for dependence, 0.81 for depression, 0.84 for legal issues, 0.82 for relational issues and 0.86 for work-related issues. Probability of event for a man between 30 and 39, drinking 2 to 4 times a week 3 or 4 glasses, never drinking more than 5 drinks at once had a risk of 13% for dependence, 6% for depression, 2% for legal issues, 4% for relational issues and 1% for work issues.  CONCLUSIONS: Alcohol related social consequences are often underestimated by drinkers. The use of predictive tools based on a limited number of variables readily available in consultation can be used by GPs to encourage occasional and regular drinkers to stop or reduce their consumption.