Purpose of review To summarize the clinical evidence on nutritional support for critically ill patients, the (patho)physiological mechanisms involved, and areas of future research. Recent findings Large randomized controlled trials have shown that early nutrition induces dose-dependent harm in critically ill patients, regardless of the feeding route, and that early high-dose amino acids are harmful. Harm has been attributed to feeding-induced suppression of cellular repair pathways including autophagy and ketogenesis, to aggravation of hyperglycemia and insulin needs, and to increased urea cycle activity. Additionally, acute critical illness was shown to be a state of anabolic resistance. The absence of benefit of early enhanced nutritional support on short- and long-term outcomes was observed in all studied subgroups. Summary While early high-dose nutrition should be avoided in all critically ill patients, the optimal initiation time of nutrition support for the individual patient, as well as ideal composition and dosing of nutrition over time remain unclear. Future studies should elucidate how fasting-induced repair pathways can be activated while avoiding prolonged starvation, and how hyperglycemia and high insulin need could be prevented. Potential strategies include intermittent fasting, ketogenic diets, ketone supplements, and alternative glucose-lowering agents, whether or not in combination with exercise.