We have investigated the association of a family history of diabetes with glucose tolerance in a population of Swedish men. All men 35-54 years of age in 1992 and living in four different local municipalities of the outer Stockholm area were screened by questionnaire. From 10236 completed questionnaires 1622 men, selected for presence of such a history but without known diabetes, as well as 1507 men without a family history underwent an oral glucose tolerance test. Diabetes (2 h-plasma glucose levels > 11.0 mmol/l) was detected in 55 and impaired glucose tolerance (plasma glucose levels 7.8-11.0 mmol/l) in 172 subjects. The odds ratio of diabetes, associated with a family history, was 4.1, confidence interval 2.1-8.3 and for impaired glucose tolerance 1.6, confidence interval 1.2-2.3. Influence of a family history was measurable also within the range of normal 2-h glucose concentrations: compared to 2-h glucose levels < 3.8 mmol/l; the odds ratio associated with a family history was 1.4, confidence interval 1.1-1.7 and 1.3, confidence interval 1.1-1.6 for concentrations 4.8-5.7 mmol/l and 5.8-7.7 mmol/l respectively. The odds ratio of diabetes and impaired glucose tolerance among men with a family history increased with number and closeness of relatives with diabetes but was not affected by the gender of the family member. Overweight (BMI > 25.0 kg/m2) increased the odds ratio of diabetes in subjects with a family history, the odds ratio being 24, confidence interval 3-177, when both conditions were present. In subjects with Type II (non-insulin-dependent) diabetes mellitus discovered during the investigation, the presence of a family history of diabetes was associated with decreased insulin secretion rather than insulin resistance as assessed by fasting insulin, homeostasis model assessment, and the 2-h insulin response to the oral glucose tolerance test. We conclude that a family history of diabetes strongly but independently of gender associates with decreased glucose tolerance. Furthermore, the results are compatible with a major role for low insulin secretion in the diabetogenic influence of a family history of diabetes in middle-aged Swedish men. Lastly, the very high risk for diabetes in middle-aged men with both a family history of diabetes and obesity indicates that such people should, for the purpose of therapeutic intervention, be identified in the general population.