Insulin lispro is absorbed more rapidly and has a shorter duration of action than regular human insulin. It improves glycaemic control but large-scale studies are required to identify regimens that optimise efficacy and safety with local dietary habits. This study involved 1184 Italian patients with Type 1 diabetes, randomised to insulin lispro (n=586) or regular human insulin (n=598) as pre-meal bolus for 3 months. Optimisation of basal NPH insulin was carried out in both groups. The number of administrations of NPH insulin was increased when using insulin lispro but, because basal and bolus insulins were mixed before meals, the total number of injections per day was unchanged. Compliance to administration time was significantly (p<0.001) greater with insulin lispro than with regular human insulin. Post-prandial blood glucose levels were lower with insulin lispro after breakfast (p<0.001), lunch (p<0.005) and dinner (p<0.001). The HbA1c level was decreased from baseline by both insulins, but the percent increase in patients with acceptable (<8%) HbA1c was greater with insulin lispro. While frequency of hypoglycaemia was decreased from baseline by both insulins, the proportion of episodes classified as severe was significantly increased from baseline with regular human insulin, but not with insulin lispro. Thus, compared with regular human insulin, improved glycaemic control was achieved with insulin lispro without an increase in severe hypoglyeaemia.