OBJECTIVE Insulin lispro used in an intensive basal/bolus regimen produces equivalent glycaemic control to human-soluble insulin but reduces rates of hypoglycaemia. We tested the hypothesis that the use of rapid-acting analogues might prevent the development of defective hypoglycaemic counterregulation during intensive insulin therapy. METHODS Ten patients with type 1 diabetes (four female, mean age 33 +/- 3 years, diabetes duration 12 +/- 2 years) participated in an open, randomized cross-over study, with 2 months run-in and 4-month treatment periods using either lispro or human-soluble insulin before meals and human NPH insulin (NPH) at night. The total of reported hypoglycaemic episodes (lispro vs. soluble, 123 vs. 128) and HbA(1c) (6.1 +/- 0.2 vs. 6.6 +/- 0.2%) were similar during both treatments. At the end of each period, we measured symptomatic, counterregulatory and cognitive responses, and glycaemic thresholds during hypoglycaemia, induced with a hyperinsulinaemic clamp (blood glucose of 5, 4.5, 3.5 and 2.5 mmol/l). RESULTS We found similar overall responses of adrenaline, cortisol, growth hormone and total symptom score. Glycaemic thresholds for rises in adrenaline (3.1 +/- 0.2 vs. 3.1 +/- 0.2 mmol/l, p = 0.76), cortisol (2.2 +/- 0.1 vs. 2.2 +/- 0.1 mmol/l, p = 0.16), growth hormone (3.3 +/- 0.15 vs. 2.9 +/- 0.2 mmol/l, p = 0.13), symptoms (3.2 +/- 0.2 vs. 3.3 +/- 0.1 mmol/l, p = 0.051) and impaired cognitive function (3.0 +/- 0.2 vs. 3.0 +/-0.2 mmol/l, p = 0.20) were also similar. CONCLUSIONS Four months of intensive treatment, with insulin lispro used pre-prandially and isophane at night, produced relatively preserved but equivalent physiological responses to hypoglycaemia as those on soluble insulin. Longer periods of treatment or alternative regimens may be necessary to demonstrate beneficial effects on hypoglycaemic physiological responses.