The incidence of both obesity and type 2 diabetes continues to rise, creating a major worldwide public health challenge. Understanding the mechanisms determining the normalisation of blood glucose levels and the limitations to complete reversal of diabetes by bariatric surgery or a very low calorie diet (VLCD) has important implications for the treatment of type 2 diabetes, but also for improving our understanding of the pathophysiology. A unifying hypothesis to explain the major pathophysiological changes in type 2 diabetes, hepatic insulin resistance and pancreatic beta cell insufficiency, involves excess triglyceride accumulation in liver and pancreas. This is supported by in vitro demonstration of impaired insulin signalling (and thereby insulin resistance) and defective insulin secretion, induced by the toxic metabolites of fat. Triglyceride content in liver and pancreas can now be measured non-invasively and precisely using the three point Dixon magnetic resonance technique. This thesis presents data on a direct comparison of bariatric surgery and VLCD, suggesting that the mechanisms involved in the reversibility of type 2 diabetes using these two interventions are similar. Pancreatic triglyceride content appears to decrease with substantial weight loss in obese individuals with type 2 diabetes but not those with normal glucose tolerance. The clinical characteristics which limit reversal of diabetes are investigated, particularly the effect of longer diabetes duration. Both a retrospective study of bariatric surgery and a prospective study using VLCD suggest that long duration diabetes is reversible, however, normal blood glucose levels are less likely to be achieved than in short duration disease. Diabetes duration may be a surrogate marker for beta cell reserve. Finally, the longer term durability of the beneficial effects of an 8 week very low calorie diet is demonstrated. Over a subsequent 6 month weight maintenance period the decrease in hepatic and pancreatic triglyceride content, and the improvements in hepatic insulin sensitivity and first phase insulin secretion are maintained. The latter appears to be a key mechanism for determining a glucose response to VLCD.