Objective To determine whether there is a longitudinal relationship between hypertension and hyperinsulinemia and to find the most useful parameter(s) for predicting the subsequent development of hypertension. Subjects and methods The oral glucose (75 g) tolerance test (OGTT) was performed in 313 patients, who were divided into three groups according to glucose tolerance based on the WHO criteria: normal, borderline and diabetes mellitus. The fasting insulin (IRI) levels, Sigma IRI (the sum of the insulin levels 0, 30, 60 and 120 min after the OGTT), insulinogenic index and Homa's index, a candidate for the simple assessment of insulin sensitivity, of the normotensive and hypertensive subjects in each subgroup were compared. In addition, 145 normotensive subjects were followed up for over 3 years and observed for the development of hypertension. Results Hypertensive diabetic subjects had not only higher fasting IRI levels and Sigma IRI values, but they also had higher Homa's indices than normotensive diabetics. Normotensive subjects with normal glucose tolerance (n = 20) did not develop hypertension. However, 16 out of 94 patients with borderline glucose tolerance and five out of 31 diabetics became hypertensive. The incidence of hypertension in the group with fasting IRI greater than or equal to 15, Sigma IRI greater than or equal to 150 or Homa's index greater than or equal to 4 was between 5 and 9 times higher than that in the group with fasting IRI <10, Sigma IRI <100 or Homa's index < 2. This difference was still significant when multivariate analysis, including various factors such as age, body mass index (BMI) and sex, was performed. Conclusions These results suggest that higher plasma IRI levels and/or insulin resistance are closely related to the pathogenesis of hypertension in patients with diabetes mellitus. Homa's index, fasting and Sigma IRI may be useful predictors of the subsequent development of hypertension. J Hypertens 2000, 18:83-88 (C) Lippincott Williams & Wilkins.