This thesis examines methods of perioperative risk stratification and outcome in patients receiving multidisciplinary stage-directed treatment for oesophagogastric cancer. The hypotheses tested were: Suboptimal bioelectrical impedance analysis (BIA) body composition variables predict poor outcomes in oesophagogastric cancer (OGC) surgery; low CT-measured psoas muscle density (PMD) predicts poor outcomes in OGC surgery; suboptimal cardiopulmonary exercise (CPX) performance predicts poor outcomes following OGC surgery; the literature offers evidence in support of enhanced recovery programmes in OGC surgery; the use of an enhanced recovery programme in OGC surgery is feasible, safe and not associated with adverse outcomes. High values for BIA-derived measures of fat-free mass and muscle mass respectively predicted longer survival (p=0.047, p=0.011), but were not associated with reduced 30-day mortality, major morbidity or length of stay. CT-measured psoas muscle density greater than the median of 48.7 Hounsfield Units predicted longer survival (p=0.046), but was not associated with reduced 30-day mortality, major morbidity or length of stay (LOHS). Multivariable analysis demonstrated radiological TNM stage (p=0.015), and both left (p=0.046) and right PMD (p=0.047), as significant and independent predictors of survival. Cardiopulmonary exercise testing results materially altered the management plan in 6.8% patients. Major morbidity (p=0.049) and poor survival (p=0.048) were associated with a high ventilatory equivalent for carbon dioxide (VE/VCO2), but not with the anaerobic threshold (AT) or peak oxygen uptake (VO2peak). VE/VCO2 also emerged on multivariable analysis as an independent and significant predictor of LOHS (p=0.001). Systematic review and meta-analysis revealed enhanced recovery programmes (ERPs) in OGC surgery to be feasible, safe and costeffective, significantly shortening length of stay (LOHS, p<0.0001). In our unit, the introduction of ERPs in gastric and oesophageal cancer surgery respectively, significantly reduced LOHS (p=0.004; p=0.032), critical care stay (p<0.0001; p<0.0001) and overall cost (p=0.001; p<0.0001).