Mortality rates from coronary heart disease and from all causes have been ascertained over ten years in three groups of people participating in the Bedford Survey--newly-diagnosed diabetics, borderline diabetics and control subjects with normal glucose tolerance. Age corrected mortality rates, from all causes and coronary heart disease, were highest in the diabetics and intermediate in the borderline diabetics and in both groups were similar in men and women. When statistical allowance was made for baseline differences in age, blood pressure and obesity, female borderline diabetics still had a significantly increased risk over their controls of death from 'all causes'. Much of the difference appeared to be due to a relative excess of deaths due to coronary heart disease. It is concluded that borderline diabetes (or impaired glucose tolerance) is associated with a relatively greater increase in mortality risk in women than men. During the 10-year follow-up of the Bedford borderline diabetics, coronary heart disease morbidity and mortality rates were similar in men and women. Age at entry to the study was the major independent and significant predictor of mortality from all causes. The level of systolic blood pressure and current cigarette smoking at baseline were statistically significant predictors only of mortality due to coronary heart disease.