Insulin is frequently required in total parenteral nutrition (TPN) solutions to control hyperglycemia. The purpose of this study was to evaluate the recovery of human insulin from standard TPN solutions with and without lipids and from TPN solutions with specialized amino acid formulations and to compare it to the insulin recovery from normal saline. All solutions were mixed in currently utilized PVC-free bags (ethylene vinyl acetate) and drained through PVC-containing tubing. Human insulin (Humulin-R) was spiked with 125I-labeled insulin and then added in concentrations of 10, 25, and 50 units to 1-liter bags containing 39-g amino acids (10% Freamine-III; or 6.9% Freamine HBC; or 8% Hepatamine), 257-g dextrose, electrolytes (Hyperlyte-R), 1000 units of heparin, MVI-12, and MTE-5 Concentrate. Alternate sets of bags contained 125 ml of 20% Intralipid and an appropriate amount of sterile water to keep the final volume at 1 liter. Actual clinical conditions of preparation, storage, and administration were simulated in this in vitro experiment. Multiple samples were collected during the 8-hr infusion period directly in gamma counter vials. All experiments and assays were done in triplicate. Our findings indicate that human insulin availability in TPN solutions is much higher (90%-95%) than the 50% suggested in the literature. Insulin recovery was not appreciably altered by adding lipids or by using Freamine HBC. Insulin recovery from TPN solutions was significantly reduced if they contained Hepatamine (87% and 88%, p less than 0.05) as compared to Freamine (90% and 94%).(ABSTRACT TRUNCATED AT 250 WORDS)