The time-action profile of the insulin analogue insulin lispro ([Lys(B28), Pro(B29)] human insulin) with its rapid onset and short duration of action might be more suitable to limit hyperglycaemic excursions after a meal rich in rapidly absorbable carbohydrates in comparison to regular human insulin. A randomized, double-blind study was performed in 10 Type I diabetic patients with good metabolic control (HbA1c 7.0 +/- 0.5%). After a baseline period of 3 h (blood glucose clamped at 6.7 mmol l-1, i.v. insulin infusion of 0.2 mU kg-1 min-1 throughout the study), the patients ate a pizza, drank a cola and had a carbohydrate-rich dessert (total carbohydrate content 140 g). Immediately before the meal 15.4 +/- 3.5 U of either insulin preparation were injected subcutaneously. Blood glucose concentrations were monitored continuously thereafter. Following the injection of insulin lispro the area under the blood glucose curve after the meal was 78% of that of regular insulin (1.76 +/- 0.34 vs 2.26 +/- 0.68 mol l-1 *240 min-1; p < 0.01). Maximal blood glucose excursions were higher and were reached later after regular insulin as compared to insulin lispro (11.9 +/- 2.8 vs 9.9 +/- 1.4 mmol l-1; p < 0.05; 66 +/- 37 vs 41 +/- 7 min; p < 0.05). Maximal individual differences in the blood glucose excursions (regular human insulin minus insulin lispro) were 4.8 +/- 2.2 mmol l-1 (p < 0.0001 against zero) after 110 +/- 37 min. In Type 1 diabetic patients prandial blood glucose excursions after a carbohydrate rich meal were reduced after preprandial injection of insulin lispro in comparison to human regular insulin.