Mortality according to glucose tolerance was studied to determine the prognosis of impaired glucose tolerance. Among 2500 persons tested in a community screening programme in 1962–1965 and followed-up for mortality to the end of 1987, age-sex-adjusted mortality rates were 37.9 ± 1.9, 53.6 ± 4.2, and 70.1 ± 3.6 deaths per 1000 person-years ( ± SE) in those with normal glucose tolerance, impaired glucose tolerance, and diabetes by World Health Organization criteria at baseline. Age-sex-adjusted mortality rates due to ischaemic heart disease were 14.3 ± 1.1, 16.3 ± 2.4, and 25.8 ± 2.0 deaths per 1000 person-years, respectively. Using criteria predating those of the World Health Organization 147 men with abnormal glucose tolerance were entered into a randomized clinical trial in which 49 were treated with tolbutamide for approximately 10 years. Those treated had lower mortality rates from all causes (mortality rate ratio = 0.66, 95 % confidence interval = 0.39, 1.10) and from ischaemic heart disease (mortality rate ratio = 0.42, 95 % confidence interval = 0.16, 1.12) than those not receiving tolbutamide. Thus mortality rates are increased in persons with impaired glucose tolerance and diabetes, and the small clinical trial suggests that tolbutamide may be beneficial in men with abnormal glucose tolerance. [Diabetologia (1997) 40: 680–686]