A prospective randomized study of the immunological effects of three total parenteral nutrition (TPN) regimens in patients undergoing preoperative parenteral nutrition was conducted. In one regimen the calories were derived solely from glucose. The others were identical except that 50 per cent of the calories were provided as lipid emulsion, in one as long-chain triglycerides (LCT) only while the other contained half the fat as medium-chain triglycerides (MCT) and half as LCT (MCT/LCT). Natural killer (NK) activity and lymphokine-activated killer (LAK) activity were significantly higher after TPN with the MCT/LCT solution. A significant fall in LAK activity occurred after TPN with the LCT solution. The interleukin 2 content in supernatants from activated T lymphocytes was significantly higher after TPN with the LCT-containing solution. Solutions containing LCT and those containing MCT perturb cytokine interactions, but this is less with MCT-containing solutions, which may augment certain responses. These observations may have implications for the design of TPN regimens.