To examine patterns of use and clinical outcomes of total parenteral nutrition (TPN). A prospective six-month audit (December 1992-June 1993). All inpatients administered TPN at a metropolitan teaching hospital during the audit period. Process measures included data about TPN initiation (bodyweight, period not receiving oral/nasogastric feeding, serum albumin level, compliance with hospital guidelines), TPN delivery data (kilojoules, and nutrient and electrolyte content), and bases for cessation or changes of TPN (biochemistry data, gastric and intestinal function). Outcome measures included body mass change, infection rate, detection of biochemical abnormalities, and death. During the audit 168 consecutive patients received 175 TPN courses. These patients were followed until discharge or death; 49 patients (29%) died. Intensive care units accounted for 57.7% of TPN use. Deviations from approved hospital guidelines for initiation of TPN were common. Only a minority of patients were malnourished on objective audit criteria; 18% of men and 13% of women were underweight by body mass index criteria and 36% were malnourished when serum albumin level (< 30 g/L) was considered. Early initiation of TPN outside accepted guidelines was common. Complications included bacteraemia (9.1% of patients tested) and catheter-tip sepsis (55.2% of 87 catheters tested). Four patients died; line sepsis caused one death and probably a further two. The incidence of glucose intolerance was 36.5%, and 25% had markers of abnormal liver function. TPN use is associated with a high risk of morbidity, and a 1.7% mortality. We recommend better patient selection for TPN, more appropriate use of enteral feeding, better infection control procedures, avoidance of substrate overload (particularly glucose), and earlier change to enteral nutrition.