Fifteen year mortality rates are reported for men participating in the Whitehall Study in 1968-70. Subjects were divided into four groups normoglycaemic (centiles 1-95 of the blood glucose distribution: n = 17,051), glucose intolerant (centiles 96-100: n = 999), newly diagnosed diabetic patients (n = 56) and previously diagnosed diabetic patients (n = 121) treated with diet +/- tablets. Relative risks for all causes mortality and from coronary and cardiovascular disease deaths were calculated. Age adjusted relative risks were highest in the newly diagnosed diabetic patients and were also increased in glucose intolerant and previously diagnosed diabetic men (p less than 0.05), but did not increase with increasing duration of diabetes. With adjustment for other risk factors, relative risks were similar in newly diagnosed and previously diagnosed diabetic men. There was no significant linear trend of adjusted relative risks with duration of diabetes when all diabetic men were pooled and person years at risk calculated. The lack of effect of duration upon relative risk together with other observations suggests common, possibly genetic, antecedents of both Type 2 (non-insulin-dependent) diabetes and coronary heart disease.