The aim of this study was to compare the causes of death and parameters related to alcohol consumption, between subjects diagnosed as diabetic, clinically by their general practitioner, or glucose intolerant and in particular as diabetic, using the epidemiological criteria of an abnormal glucose level following an oral glucose tolerance test. The subjects in this study were 7035 working men, aged between 44 and 55 years, who attended the first follow-up examination of the Paris Prospective Study, between 1968 and 1973. They were classified as 'clinically diagnosed diabetic' or, following an oral glucose tolerance test and the World Health Organisation criteria, as having 'oral glucose tolerance test diagnosed diabetes', impaired glucose tolerance or normoglycaemia. The relative risk of death by cirrhosis, in comparison with the normoglycaemic group, was 21 (95% confidence interval: 9.1-49) in the group diagnosed diabetic by the oral glucose tolerance test, significantly different (p less than 0.02) from the group diagnosed diabetic clinically 3.1 (0.41-24); factors indicative of excessive alcohol consumption at baseline differed accordingly. In contrast, the relative risks for death by coronary heart disease were similar, 2.1 (1.0-4.1) and 2.7 (1.4-5.4) respectively; all of the factors defining the insulin resistance 'Syndrome X' (hyperglycaemia, hyperinsulinaemia, hypertension, hyperlipidaemia and also central obesity) and predictive of coronary heart disease were elevated in both groups of diabetic subjects. 'Diabetes', as diagnosed by the oral glucose tolerance test, might be the consequence of excessive alcohol consumption which could lead to insulin resistance, then to coronary heart disease, as well as to alcohol-related diseases.