Critical illness affects ~130,000 Australians each year, costing the health-care system nearly $3 billion. For intensive care unit (ICU) survivors, quality of life and functional recovery are compromised, with symptoms persisting five years post-discharge. Patients admitted to ICU with traumatic brain injury (TBI) are at particular risk. Accordingly, interventions that enhance recovery will improve patients' quality of life and are also likely to be cost-effective. Nutrition therapy, ingested or delivered artificially, is an essential component of clinical practice in ICU and post-ICU. In this thesis I reviewed the extent of nutrition research in a hospitalised TBI population (Chapter 1) to establish insufficient data reporting intake post-ICU. In heterogeneous cohorts of critically ill patients, nutrient delivery during ICU admission is below prescribed targets. From a large international cohort, I determined that energy and protein delivery to ICU patients with TBI is below targets, and deficits in the first 12 days are associated with longer time to discharge alive from ICU and hospital, and prolonged mechanical ventilation (Chapter 4). In a methodologically-rigorous single-centre observational study I established that energy and protein deficits exist in ICU. Perhaps of more concern, these deficits increase post-ICU leading to cumulative deficits throughout hospitalisation (Chapter 1). These observations highlighted methodological issues, particularly with weighed food records to measure oral intake of hospitalised individuals (Chapter 2). Logistical and attitudinal barriers impede nutrition delivery. Interviews with medical and nursing practitioners provided insight into why these occur (Chapter 1). Additionally, TBI patients have marked changes in ultrasound-derived quadriceps muscle thickness. I established that this novel methodology, while challenging, is feasible and may correlate with total lean mass and long-term function (Chapter 3). To provide context beyond the cohort of TBI patients I explored relationships between nutritional intake during critical illness and long-term function. In a blinded pilot trial of critically ill patients, those randomised to augmented enteral nutrition to deliver greater energy, were more likely to return to work after 12- months than those receiving standard nutrition (Chapter 4). In addition, there is considerable interest within the critical care community on the effect of protein delivery on outcomes. I conducted a meta-analysis of randomised controlled trials (RCTs) with greater or lesser amounts of protein delivered to critically ill patients and did not observe any effect of greater protein dose on clinical outcomes. However even the cohort receiving greater protein had amounts lower than recommended in international guidelines. Lastly, because a frequent criticism of the role of nutritional therapy in the critically ill is the lack of effect on mortality, I undertook a systematic review and identified that nutrition intervention studies in critical care with the primary outcome of mortality have utilised sample size calculations that require a large, and possibly implausible, effect on mortality. The implications are that investigators should incorporate more realistic estimates of effect size in the future and that previous RCTs may have failed to detect an effect on mortality even if there was such an effect (Chapter 5).