An association between insulin resistance and hypertension has been reported in several studies. In apparent contradiction, insulin infusion in healthy volunteers is associated with vasodilatation. Furthermore, there is evidence that some insulin effects may differ between the sexes. We performed three-step hyperinsulinaemic-euglycaemic clamp studies in six men and six women to test the hypotheses that: 1) insulin might affect the release of vasoactive substances by the endothelium, and 2): this putative effect on vasoactive substances might differ between men and women. Six other women and six men served as control subjects, receiving 154 mmol/l NaCl (saline) infusion. Plasma levels of insulin, immunoreactive endothelin, L-arginine (precursor of nitric oxide), L-citrulline (by-product of nitric oxide synthesis) and cyclic GMP (second messenger of nitric oxide) were measured during infusion of insulin or 154 mmol/l NaCl (saline), respectively. We also assessed urinary excretion of 6-keto PGF-1 alpha (a degradation product of prostacyclin reflecting prostacyclin production). Blood pressure was monitored in all subjects throughout the experiment. In women plasma levels of immunoreactive endothelin decreased from (mean +/- SD) 2.58 +/- 0.96 to 1.7 +/- 0.72 pmol/l during insulin infusion (p < 0.01), while remaining constant in female control subjects (p < 0.02). No changes in levels of endothelin were observed in men during infusion of insulin or saline. In women levels of cGMP rose and levels of L-arginine decreased significantly during insulin infusion, consistent with an increase in nitric oxide production. Excretion of 6-keto PGF-1 alpha also increased significantly in women during insulin infusion. No such effects were observed in men, or in women during infusion of saline. Blood pressure remained constant in all subjects during hyperinsulinaemia. We conclude that sex differences exist in the effects of insulin on the endothelium. Short-term hyperinsulinaemia in women is associated with a decline in levels of immunoreactive endothelin, and possibly with a rise in production of nitric oxide and prostacyclin. In contrast, levels of vasoactive substances remained constant in men during hyperinsulinaemia. Our findings may partly explain insulin's vasodilatory effects in healthy individuals. It remains to be investigated whether these effects are lost in insulin-resistant states. Our observation that there is a sex difference in insulin effects on the endothelium may help explain why the link between hyperinsulinaemia and cardiovascular disease appears to be clearer in men than in women.