Some physicochemical factors that could account for insulin adherence to type I glass bottles from admixtures of insulin with 5% dextrose (D5W) and 0.9% sodium chloride (NS) injections were studied. Samples of three volumes of NS or D5W containing insulin I 125 were mixed in three sizes of bottles to test the effect of surface area and volume. Appropriate volumes of insulin were combined with insulin I 125 and D5W or NS to yield solutions containing nine concentrations of insulin to test the effect of insulin concentration. Appropriate volumes of KCl injection to yield six concentrations were combined with insulin I 125 and NS or D5W to test the effect of KCl concentration. All samples were assayed by gamma scintillation. In general, there was a direct relationship between the percentage of insulin adhering and the container surface area. In D5W admixtures, as the fill volume at constant insulin I 125 concentration was doubled and quadrupled, the adherence of insulin decreased in all three bottle sizes (200, 250, and 500 ml). In NS admixtures, however, this effect was seen only with the 250-ml bottle. Increasing insulin concentrations over the range of 50-300 units/liter in D5W and 0-50 units/liter in NS resulted in decreased adherence. The addition of 1-60 meq/liter of KCl resulted in a significant decrease of insulin adhering from D5W and an insignificant decrease from NS admixtures. The percentage of insulin adhering to type I glass surfaces may be reduced to 25% or less by preparing i.v. admixtures in full bottles of D5W or NS at insulin concentrations of 25 to 300 units/liter. The addition of KCl, when therapeutically appropriate, will further decrease the extent of insulin adherence.