The evidence linking insulin to blood pressure is controversial, and results for groups similarly categorized by body mass, glucose tolerance and hypertensive status are often contradictory. We have investigated the relationship in three population-based samples of Micronesian (Nauru), Polynesian (Western Samoa) and Melanesian (New Caledonia) Pacific islanders, who are known to be susceptible to obesity, non-insulin-dependent diabetes mellitus (NIDDM), hyperinsulinaemia and hypertension. After controlling for age and body mass index (BMI), mean fasting and 2-hr (post 75 g glucose) insulin levels were not significantly different between hypertensive and non-hypertensive subjects, in any population or glucose tolerance sub-group, excepting 2-hr insulin in New Caledonians with normal glucose tolerance. Similarly, there were no strong trends for a higher prevalence of hypertension amongst those with insulin levels in the upper quartile of the distribution of each population, although it was apparent in some BMI/glucose tolerance sub-groups of two populations. Multiple linear regression analyses also showed an inconsistent and where present, weak, independent association between insulin and blood pressure in models predicting both systolic and diastolic blood pressure. We therefore conclude that the hypothesis implicating insulin as a major determinant of blood pressure and as the pathophysiological link between obesity, NIDDM and hypertension is not strongly supported either by the literature or the present data.