This review analyses the recently published literature focusing on nutritional assessment in intensive care unit patients. The metabolic response to nutritional intervention is difficult to evaluate in critically ill patients whose body weight is influenced largely by massive fluid administration or losses. Visceral protein plasma levels have been proposed for this purpose, because they reflect hepatic synthesis in response to nutrient supply. However, in acute inflammatory states, liver activity is converted to the synthesis of acute-phase response proteins, resulting in a dramatic drop in visceral proteins, despite nutritional support. The data regarding visceral protein levels were examined in relation to nutritional supplementation, and compared with other nutritional parameters and clinical outcomes. Transthyretin and retinol-binding protein levels seem to be the most sensitive to nutritional intervention. They are also the earliest to rise at the decrease of acute-phase protein levels, therefore representing a good index of the reversing reprioritization of hepatic protein synthesis. An inconsistent relationship was found between visceral protein plasma levels and clinical outcome in intensive care unit patients, probably because of the difficulty in demonstrating clearly a beneficial effect of nutritional supplementation in highly catabolic conditions. In the acute stage of critical illness, the bi-weekly measurement of transthyretin together with acute-phase response protein plasma levels seems to be a 'window' on the metabolic condition (anabolism versus catabolism). However, only in the presence of stable inflammatory parameters do transthyretin levels reflect the adequacy of nutritional coverage.