The difficulties of achieving good glycaemic control in insulin-dependent diabetes are due in large part to the inadequacies of subcutaneously administered insulin. In particular, resorption with the long acting form is variable from one subject to another and from one day to another and irregular over time, whereas the action of the rapid acting form is too late and prolonged. The slowness of absorption of the rapid acting form is attributable to the need for hexamer dissociation. The Lilly Laboratories, by inverting the amino acids lysine and proline in positions 28 and 29 in the B chain, have created an insulin (Lispro) which more rapidly dissociates into monomers after injection. The stability of Lispro is good, probably because of its phosphate buffer. In our experience, in conditions simulating use in portable insulin pumps, Lispro proved to be more stable than insulins specially intended for this use. The affinity in vitro was identical to that of insulin for its receptor. The affinity for insulin-like growth factor-I (IGF-I) receptor has been found to be 1.5 times as high as that of fast insulin in some models and comparable in others, and nearly 1,000 times less than that of IGF-1. Studies on in vivo potency and ex vivo cell growth, as well as of tolerance in the animal (mutagenicity, toxicity and carcinogenicity), have not shown a different effect from regular insulin (contrary to results for analogue Asp B10). The pharmacokinetics has shown an earlier and higher insulinaemic peak and a more rapid return to baseline values than regular insulin. On the basis of pharmacokinetic studies in normal subjects and diabetic patients, the characteristics retained by the licensing authorities are onset of action at 15 min, insulinemic peak between 30 and 70 min, and duration of action 2 to 5 h. Some clinical studies have shown a shortened action period of 1 to 3 h as compared to regular insulin and less influence of dose and injection site, notably with a return to normal insulin levels. The time required for normalisation is increased by 1 h if the injection is made in the thigh rather than the abdomen, as compared to 2 to 3 h for conventional insulin. This suggests that Lispro should be administered just before the meal (0 to 15 min). In some patients, an insulin with prolonged action can be added if the interval between injections is prolonged, i.e. always at the evening meal but possibly also at the noon meal. Lispro can be mixed with Umuline NPH or Umuline zinc without any alteration in its pharmacokinetics and potency if the injection is performed immediately. The few studies that have considered glycaemic stability and reproducibility have shown a tendency toward improvement in glycaemic excursions during the day, as measured by MAGE, and in insulinaemia variability expressed in area under the curve, which was reduced by half in the same individual or from one individual to another, with less marked impact on the variability from one day to another of glycaemic excursions. On the whole, Lispro provides faster kinetics, greater stability and possibly better reproducibility than fast insulin. These advantages, if confirmed by clinical experience, should allow an improvement in the comfort and glycaemic stability of diabetic patients.