A pig model of insulin-dependent diabetes was used to examine the importance of the portal-systemic insulin gradient for whole-body metabolic control. Six pigs had jugular vein, portal vein, and carotid artery cannulae implanted before being made diabetic (150 mg kg[-1] streptozotocin). Each animal received 4 weeks of portal and 4 weeks of peripheral insulin delivery in random order. The blood glucose target range was 5-10 mmol x l(-1), and serum fructosamine and fasting and postprandial blood glucose concentrations were not different between peripheral and portal insulin infusion. Insulin requirement was not different between the 4 week infusion periods, but fasting peripheral insulin levels after peripheral delivery (124 +/- 16 (mean +/- SEM) pmol x l[-1]) were significantly higher (p < 0.05) than in portally infused (73.8 +/- 5.4 pmol x l[-1]) or pre-diabetic control animals (68.4 +/- 3.6 pmol x l[-1]). Basal hepatic glucose output was also higher (p < 0.05) in peripherally (4.2 +/- 0.4 mg x kg[-1] x min[-1]) than in portally infused animals (2.9 +/- 0.4 mg x kg[-1] x min[-1]) or controls (3.0 +/- 0.3 mg x kg[-1] x min[-1]). Clamp glucose metabolic clearance rate was, however, not different between the peripheral and portal insulin delivery routes (8.1 +/- 1.0 vs 9.0 +/- 0.7 ml x kg[-1] x min[-1]), although both were significantly lower (p < 0.05) than that measured in prediabetic control animals (11.7 +/- 1.0 ml x kg[-1] x min[-1]). Lipid profiles and subfractions were similar in all three groups. It is concluded that the portal route of delivery is superior to the peripheral in maintaining more appropriate insulin concentrations and control of hepatic glucose output, although in the absence of euglycaemia it is still associated with significant metabolic abnormalities.