A longitudinal population study of 1462 women, aged 38-60, was carried out in Gothenburg, Sweden in 1968-69. Women with initially manifest diabetes mellitus had significantly increased 12-year incidence of myocardial infarction and increased mortality while no increased incidence of angina pectoris, ECG changes indicating ischaemic heart disease or stroke was observed. The association to myocardial infarction remained in multivariate analyses and was independent of age, body fat distribution, smoking, serum cholesterol and systolic blood pressure. The association to mortality was independent of these factors and also of serum triglycerides. Women who were diagnosed as "new diabetics" during the 12-year follow-up had a significantly increased 12-year incidence of myocardial infarction but no significant increase was observed for any of the other end-points studied. When women with initially manifest diabetes mellitus were excluded, no association was found between initial fasting blood glucose concentration and the end-points studied. A negative significant association was found between initial fasting blood glucose concentration and smoking.