A longitudinal population study of 1462 women aged 38-60 was carried out from 1968-9 to 1980-1 in Gothenburg, Sweden. The initial and follow up examinations included questions concerning history of diabetes and antihypertensive treatment. A considerably increased risk of developing diabetes was observed for subjects with hypertension taking diuretics (895 patient years), subjects taking beta blockers (682 patient years), and subjects taking a combination of diuretics and beta blockers (281 patient years) compared with subjects not taking antihypertensive drugs (13 855 control years). When diuretics and beta blockers were compared no difference was found in relative risk. Despite this increased risk, and because little is known about the relation between other forms of antihypertensive treatment and diabetes, diuretics and beta blockers should remain the treatments of choice in arterial hypertension.