Insulin lispro was compared with regular human insulin with respect to glycaemic control in patients with diabetes mellitus on intensive insulin treatment. Sixty-two patients (55 type 1; 7 type 2) from eight study centres in the Czech Republic, Slovenia and the Slovak Republic participated in a 4-month, open-label, randomized, crossover study. Patients administered insulin lispro immediately before meals or regular human insulin 30 min before meals. A test meal (220-400 kcal), based on local and individual dietary habits and consistent for each patient throughout the study was given at baseline and at the end of each treatment. At each test meal visit HbA1c, fasting blood glucose, 1-hour and 2-hour postprandial blood glucose levels were measured. The level of HbA1c (7.6% +/- 1.5% versus 7.4% +/- 1.5%), incidence of hypoglycaemia (41-66% of patients--versus 39-63%) and daily insulin dose (0.67 +/- 0.11 U/kg versus 0.65 +/- 0.11 U/kg) did not differ between treatment groups at endpoint (insulin lispro versus regular human insulin, respectively). Mean 2-hour postprandial blood glucose excursion for the insulin lispro group (0.0 +/- 3.7 mmol/L) was significantly lower (p = 0.035) when compared with the regular human insulin group (1.3 +/- 3.7 mmol/L) at endpoint. Therapy with insulin lispro was therefore associated with a significant improvement in postprandial blood glucose excursion control when compared with regular human insulin, without an increase in rate of hypoglycaemia.