Short-term exposure of tissues to pulses of insulin generally leads to an enhancement of insulin action. We have investigated the possible beneficial effects of long-term near-physiological continuous vs pulsatile intravenous insulin treatment of insulin-deficient streptozotocin (70 mg/kg) diabetic rats on blood glucose control, in vivo insulin action and in vitro insulin action in isolated adipocytes. First, we determined the 24-h peripheral plasma insulin profiles in normal rats under precisely controlled mealfeeding conditions. Basal plasma insulin levels (40 +/- 9 microU/ml) oscillate with a periodicity of 11.9 +/- 0.9 min (p < 0.05), and an amplitude of 60 +/- 10%. Subsequently, the 24-h insulin profile was mimicked in diabetic (D) rats by a continuous (c) or pulsatile (p) (6-min double, 6-min off) insulin infusion rate for 2 weeks, using a programmable pumpswivel unit. Control (C) rats received vehicle treatment. In Cc, Dc, Cp and Dp daily urinary glucose loss and average plasma glucose levels were 0 +/- 0, 7.5 +/- 4.4, 0 +/- 0, 0.8 +/- 0.4 mmol and 6.7 +/- 0.2, 11.5 +/- 2.7, 6.6 +/- 0.1, 5.9 +/- 1.4 mmol/l, respectively. Hypoglycaemia (< 3 mmol/l) was observed in 10 and 20% of the blood samples collected from Dc and Dp rats, respectively. After 2 weeks of treatment, in vivo peripheral and hepatic insulin action was measured by the hyperinsulinaemic euglycaemic (6 mmol/l) clamp with [3-3H]-glucose infusion. Pre-clamp counter-regulatory hormone levels were similar among rats. Compared to Cc and Cp, Dc showed a reduction in insulin sensitivity and responsiveness for peripheral glucose uptake whereas Dp only showed a reduction in insulin sensitivity. Suppression of hepatic glucose production by insulin was similar among rats. After 2.5 weeks of treatment, epididymal adipocytes were isolated. Specific [125I]-insulin binding, basal and insulin-stimulated [U-14C]-glucose uptake and isoproterenol-stimulated glycerol output were comparable among rat adipocytes. The inhibition of glycerol output by insulin was identical in Cp and Dp (V(max) = 48.6 +/- 6.1 and 42.3 +/- 4.6%) but blunted in Dc vs Cc (V(max) = 8.2 +/- 4.6 vs 44.0 +/- 7.2%, p < 0.01) adipocytes, suggesting a post-binding defect in the antilipolytic action of insulin in Dc rats. In conclusion, long-term near-physiological pulsatile intravenous insulin replacement in insulin-deficient diabetic rats is more efficient than continuous delivery in reducing blood glucose, lowering glucosuria, increasing insulin sensitivity and inhibiting lipolysis.