1. Our objective was to compare the effect of a long-acting calcium antagonist (nisoldipine) compared with an angiotensin-converting enzyme inhibitor (lisinopril) on the non-neurogenic regulation of the microvascular blood flow in hypertensive Type I diabetes patients with diabetic nephropathy. 2. We performed a 1-year double-blind, double-dummy randomized controlled study comparing nisoldipine (20-40 mg once daily) with lisinopril (10-20 mg once daily) in 48 hypertensive Type I diabetes patients with diabetic nephropathy. For comparison, 22 age-matched normotensive healthy control subjects were included. Measurements were performed at baseline and after 1 year of antihypertensive treatment. The minimal vascular resistance and distensibility (stiffness) of resistance vessels in skin and skeletal muscle were measured using the local isotope washout method. 3. Mean arterial pressure was reduced to the same extent in both groups: nisoldipine, 113 ± 2.1 to 105 ± 1.6 mmHg (P < 0.001); lisinopril, 110 ± 2.7 to 101 ± 2.1 mmHg (P < 0.002) (controls, 88 ± 2.2 mmHg; P < 0.0001 compared with diabetic patients). Nisoldipine improved the skin vascular distensibility from 28 ± 3.3 to 43 ± 3.8% (P < 0.005) and decreased skin minimal vascular resistance from 16.9 ± 1.0 to 13.6 ± 0.8 mmHg·ml-1·min·100 g (P < 0.02). Lisinopril had no significant effect on skin vascular distensibility (40 ± 4.0% and 41 ± 4.4%), but minimal vascular resistance tended to diminish (18.1 ± 0.9 to 15.8 ± 1.3 mmHg·ml-1 min 100 g (P = 0.09). Nisoldipine significantly increased the skin distensibility (P = 0.05) after 1 year of antihypertensive treatment compared with lisinopril. 4. The control group had a skin vascular distensibility of 54 ± 3.2% and a minimal vascular resistance of 10.8 ± 0.7 mmHg·ml-1·min·100 g, both significantly different from the values in the diabetic groups (P < 0.0001 for all). Skeletal muscle vascular distensibility was unaltered after 1 year of treatment with both nisoldipine (22 ± 3.3% and 19 ± 2.7%) and lisinopril (19 ± 2.1% and 24 ± 2.5%), but was reduced compared with a control value of 43 ± 3.7% (P < 0.0001 for diabetes patients versus controls). However, neither nisoldipine nor lisinopril had any effect on the increased minimal vascular resistance or the reduced skeletal muscle distensibility. 5. Enhanced thickening of the basement membranes of the terminal arteriolar wall was found in skin biopsy specimens in 91% of diabetic patients and 38% only in control subjects (P < 0.000001 both before and after treatment for diabetic patients versus controls). There was no significant effect of antihypertensive treatment on arteriolar hyalinosis. 6. The reduction in systemic blood pressure was identical during 1 year of treatment with nisoldipine or lisinopril. The abnormal arteriolar stiffness was more pronounced in the group treated with nisoldipine than with lisinopril and only nisoldipine compared with lisinopril improved the abnormal arteriolar stiffness and minimal vascular resistance in the skin. This suggests that nisoldipine can reverse the peripheral skin perfusion and thereby improve the local protection against development of ischaemic skin lesions in Type I diabetes patients with clinical diabetic nephropathy.