1. Previous studies have shown that endothelium-dependent relaxation in the aorta of spontaneously diabetic bio bred rats (BB) is impaired. 2. We have investigated noradrenaline (NA) contractility, endothelium-dependent acetylcholine (ACh) and bradykinin (BK) relaxation, and endothelium-independent sodium nitroprusside (SNP) relaxation in mesenteric resistance arteries of recent onset BB rats and established insulin treated BB rats, compared to their age-matched non diabetic controls. 3. There was no significant difference in the maximum contractile response or sensitivity to noradrenaline in either of the diabetic groups compared to their age-matched controls. 4. Incubation with the nitric oxide synthetase inhibitor NG-nitro-L-arginine (L-NOARG) resulted in a significant increase in maximum contractile response to noradrenaline in the recent onset age-matched control group (P < 0.05). Analysis of the whole dose-response curve (using ANOVA for repeated measures with paired t test) showed a significant left-ward shift following the addition of L-NOARG (P < 0.001). A similar but less marked shift (P < 0.01) was evident in vessels from recent onset diabetics. An overall shift in both sensitivity and maximum response was also evident in the age-matched non diabetic controls of the insulin-treated group (P < 0.05). However, by contrast, there was no significant change in sensitivity in the insulin-treated diabetic rats. 5. ACh-induced endothelium-dependent relaxation was significantly impaired in the recent onset diabetic rats compared to their age-matched controls (47 +/- 11% versus 92 +/- 2%, P < 0.05, n = 6), and in the insulin treated diabetic rats (34 +/- 5% versus 75 +/- 6%, P < 0.05, n = 6). The relaxation responses to BK also were significantly impaired in the diabetic rats compared to their age-matched controls (recent onset: 20 +/- 3% versus 72 +/- 7%, P < 0.05, n = 6; insulin treated: 12 +/- 9% versus 68 +/- 7%, P < 0.05, n = 7). 6. Incubation with either the nitric oxide synthetase substrate, U-arginine, or the free radical scavenging enzyme superoxide dismutase (150 mu ml-1) failed to improve the attenuated response of acetylcholine-induced relaxation in the diabetic vessels. 7. Endothelium-dependent relaxation mediated by ACh and BK was significantly attenuated in both the diabetic and control vessels after incubation with L-NOARG. 8. Pretreatment with a cyclo-oxygenase inhibitor, indomethacin, significantly enhanced the relaxation to ACh in both the recent onset and insulin treated diabetic rats (42 +/- 10%, n = 7 versus 64 +/- 7%, n = 7, P < 0.05, and 40 +/- 5%, n = 7 versus 65 +/- 9%, n = 6, P < 0.05). 9. Following endothelium removal, there was a marked impairment in endothelium-dependent relaxation responses to ACh and BK in both the diabetic and control vessels. 10. Incubation with the thromboxane A2 receptor antagonist SQ29548, did not significantly improve the ACh endothelium-dependent relaxation response in the diabetic vessels. 11. Endothelium-independent relaxation to sodium nitroprusside was significantly impaired in the first group of diabetic vessels studied; however, subsequent studies showed no impairment of the sodium nitroprusside response in the diabetic vessels. 12. In conclusion, the ability of the endothelium to regulate vascular contractility is reduced in recent onset diabetic vessels, and significantly impaired in established insulin treated diabetics. Relaxation to the endothelium-dependent vasodilators ACh and BK was impaired in both the recent onset and the established insulin treated diabetics, and the ACh response was significantly improved following pretreatment with indomethacin, suggesting a role for a cyclo-oxygenase-derived vasoconstrictor. Preliminary studies with a thromboxane A2, receptor antagonist, SQ29548 did not significantly improve the impaired relaxation to ACh, indicating that the vasoconstrictor prostanoid is not thromboxane A2.