Three hundred patients undergoing major general surgical procedures were randomized by means of a computer-assisted algorithm to receive either total parenteral nutrition (TPN) from the first postoperative day or only prolonged glucose administration (250-300 g/day) up to 15 days after operation. All patients receiving TPN were treated individually based on daily measurements of energy and nitrogen balances. The treatment goal was to keep the patients in positive energy balance (+20%) and close to nitrogen balance. The effects of the two "nutrition regimens" on outcome such as mortality rate, complications, the need of additional medical support and patient-related functional disabilities were investigated. No selection of patients was made, that is, malnourished patients were also randomized. There were no differences among TPN versus glucose treatment when results were analyzed according to intent to treat. Approximately 60% of all patients were able to start eating within 8 to 9 days after operation. No differences were observed between such patients regardless of being treated with TPN or glucose only. Patients on glucose treatment during 14 days had a significantly higher mortality rate (p < 0.05) than patients on either continuous and uncomplicated TPN treatment or short-term glucose treatment. Similar results for mortality rates also were seen with regard to severe complications (cardiopulmonary problems, sepsis, and wound-healing insufficiencies), functional disturbances, the need of additional medical support, and abnormalities in nutritional state. Twenty per cent of the patients randomized to TPN treatment showed a statistical trend (p < 0.10) toward a higher mortality rate (36%) compared with patients randomized to prolonged glucose treatment (21% mortality rate). These patients could not be identified by evaluation of preoperative factors. Thus, the overall evaluation of the results makes it likely that a fraction of high-risk patients (approximately 20%) were not doing well on immediate postoperative intravenous feeding, and it is possible that TPN to such patients accentuated their morbidity rate. Although patients (20%) on prolonged semi-starvation (14 days glucose treatment) had increased mortality rate and severe complications, it was possible that undernutrition induced a slightly different complication scenario than induced by TPN in the high-risk patients. The results demonstrate that in most surgical patients (60%), postoperative semi-starvation is not a limiting factor for outcome. In remaining 40%, inadequate nutrition was associated with both increased morbidity and mortality rates. In this sense, inadequate nutrition represents both too much and too little, whereas overfeeding seemed to be a larger problem than underfeeding.(ABSTRACT TRUNCATED AT 400 WORDS)