The association between insulin and hypertension remains equivocal. We therefore investigated insulin levels in 3037 normotensive and 1067 hypertensive subjects from the National Heart, Lung and Blood Institute (NHLBI) Family Heart Study (FHS) by two different approaches. First, we compared insulin levels between normotensive and 275 untreated hypertensive subjects. Insulin levels unadjusted as well as adjusted for age, sex, and center were significantly higher in hypertensives. After adjustment for body mass index (BMI), insulin remained significantly higher only in the diastolic hypertensive group (mean +/- SD 77.0 +/-36.7 pmol/L, P < .01) but not in the isolated systolic hypertensive group (67.0 +/- 38.2 pmol/L) when compared to normotensives (63.2 +/- 29.1 pmol/L). A sibpair analysis was then used that compared the intra-sibpair differences in insulin concentrations to the intra-sibpair differences in blood pressure (BP) levels. This approach was intended to control for the effects of genetic and residual shared environmental variance upon insulin levels. The intra-sibpair difference in insulin concentrations between concordant (diastolic and systolic deltaBP < 5 mm Hg) and discordant sibpairs (diastolic and systolic deltaBP > 15 and > 20 mm Hg, respectively) was no longer significantly different when adjusted for BMI (2.7 v 5.9 pmol/L for diastolic and -1.7 v -1.8 pmol/L for systolic BP). Even the random selection of one sibpair from each of the 326 families independently of insulin and BP levels did not result in a significant correlation between the intrasibpair differences in insulin and BP. Using an insulin resistance index instead of insulin did not change our findings. Our investigation in the FHS sample of families suggests that there is only a small, if any, influence of insulin levels on BP after adjustment for obesity-related sources of variation.