A previous study has shown that malnourished, clinically stable patients with liver cirrhosis are in protein and energy balance at their spontaneous dietary intake and that an improvement in nutritional status cannot be anticipated at this intake (Nielsen et al. 1993). In the present study we examined to what extent oral intake could be increased by nutritional support, and to what extent dietary protein would be retained with increased intake. The techniques used for balance studies were also validated since this information is not available for patients with liver cirrhosis. Fifteen malnourished patients with alcoholic liver cirrhosis were given increasing amounts of a balanced ordinary diet for 38 (SE 3) d. Intakes of protein and energy were recorded by weighing servings and leftovers on food trays. Protein intake was calculated from food tables. Total N disposal was calculated after measurement of urinary N excretion, and protein balance was calculated from the N balance. A validation study of protein balance in a subgroup of patients (analysis of N in food by the duplicate portion technique, correction for incomplete recovery of urine by measurement of urinary para-aminobenzoic acid (PABA) after administration of PABA tablets, and measurement of faecal N) did not change protein balance values. Protein intake increased from 1. 0 (SE 0.1) g/kg per d to 1·8 (SE 0·1)g/kg per d. With increasing protein intake, 84 (SE 8)% of the increase in intake was retained. The rate of protein retention was not saturated at the intakes obtained in this study. Protein intolerance was only encountered in one patient. Available evidence indicates that the requirement for achieving N balance is increased in these patients but protein retention is highly efficient with increased intake. Protein retention is dependent on energy balance. Energy intake was calculated from food tables and total energy expenditure was calculated by the factorial method. A validation study was performed in a subgroup of patients. The energy contents of food sampled by the duplicate portion technique, and of urine and faeces were measured by bomb calorimetry. Resting energy expenditure (REE) was measured by indirect calorimetry before and at the end of the study, and O2 uptake during bicycle exercise was measured before and at the end of the study. The measured intake of metabolizable energy was on average 13% lower than the value given in food tables. Calculated energy expenditure was not changed by the validation study. Mean energy intake was 163 (SE 10) kJ/kg per d and mean energy expenditure was 134 (SE 5) kJ/kg per d (P = 0·007), indicating that the protein retention described occurred at a positive energy balance. It is concluded that a substantial retention of dietary protein can be obtained by oral nutrition support over a prolonged period of time in patients with liver cirrhosis. Requirements of protein for maintenance and repletion in these patients are discussed.