A study was performed to estimate the absorption kinetics of insulin infused subcutaneously. Four insulin-dependent diabetic subjects had their insulin pumped through a subcutaneously implanted fine polyethylene catheter at a constant rate of 5.0 ± 0.3 ml/h but at two different insulin concentrations: 218 mU/ml between meals, and 2400 mU/ml at the start of breakfast, lunch, and dinner (lasting 20, 30, and 30 min, respectively). The amount (40 U/day) and distribution of insulin delivered was identical in the four patients in order to facilitate comparison between the subjects. No attempt was made to normalize their blood glucose during the study period. A study of the kinetics of insulin absorption was made by assaying plasma insulin levels; lack of plasma anti-insulin antibodies was verified; plasma C-peptide levels were measured and were far below values observed in the fed state in nondiabetic patients. The mean maximum insulin level reached after switching from low to high concentration insulin was observed 87 ± 2 min after breakfast, 117 ± 22 min after lunch, and 125 ± 20 min after dinner. Differences observed are not significant. These values are similar to those observed after subcutaneous injection of 40 U/ml Regular insulin as a single bolus. After switching from high to low concentration, plasma insulin levels did not return to their basal values before the third or fourth hour. Subcutaneous insulin infusion could be a safe and easy way of insulin administration in an open-loop system; however, this method does not seem to be suitable for a closed-loop system.