To compare daily blood glucose and free-insulin profiles after the administration of two different mixtures of semisynthetic human insulins (soluble + lente or soluble + NPH), 8 well-controlled insulin-dependent diabetic patients with identical basal blood glucose levels were studied. At the beginning of the first daily period of study, each subject was connected to an Artificial Pancreas to achieve a standard blood glucose (100 mg/dl); after at least 4 hours of normoglycemia, feed-back control was interrupted and the insulin mixture injected, at 8 p.m. and 8 a.m., at the dose previously optimised but always in the ratio 1:2 (1/3 rapid-acting + 2/3 intermediate-acting insulin). Blood glucose and free-insulin were then monitored for a 24-hour period, starting from 8 p.m. The same observation was repeated for each patient a few days later, only changing the intermediate-acting insulin. After dinner (8.30 p.m.) and after breakfast (8.30 a.m.) mean blood glucose was lower with soluble + NPH mixture, while early in the morning and in the late afternoon better glycaemic control was achieved with soluble + lente. F-IRI showed an almost specular behaviour: lower values soon after mixture injection and higher levels 8-12 hours later were observed with soluble + lente than with soluble + NPH. These data indicate that, at the ratio 1:2, the mixture of short-acting with lente HM insulin differs from the one with NPH not only for a lower activity of soluble insulin but also for a stronger and more prolonged effect of the intermediate preparation. This difference should be carefully taken into account in clinical practice.