The 15 year mortality rates and causes of death are reported for the Paris Prospective Study cohort of 7180 working men, aged between 44 and 55 years who attended the first follow-up examination. All subjects were classified as normoglycaemic, impaired glucose tolerant or diabetic according to the WHO criteria, following an oral glucose tolerance test. The relative risks of death in comparison to the normoglycaemic group were 1.6 for impaired glucose tolerant and 2.3 for diabetic subjects; for death due to coronary heart disease: 1.7 and 2.3 respectively; for death due to alcohol and cirrhosis: 7.0 and 13.3 respectively. It appears that in this cohort a proportion of subjects screened as diabetic have impaired liver function and disturbed carbohydrate metabolism, due to excessive alcohol consumption. Alcohol should be investigated as a possible risk factor for diabetes, particularly in epidemiological studies where diabetes is diagnosed by the oral glucose tolerance test and the population has a high consumption of alcohol.