Low-carbohydrate diets are increasingly used to help patients with obesity and type 2 diabetes. We sought to provide an overview of the evidence for this treatment approach, considering the epidemiology and pathophysiology of obesity and diabetes in terms of carbohydrate excess. We describe the mechanistic basis for the clinical benefits associated with nutritional ketosis and identify areas of practice where the evidence base could be improved. We summarize the key principles which inform our approach to treating patients with low-carbohydrate diets. The scientific controversy relating to these diets is real but is consistent with the known challenges of any dietary interventions and also the limitations of nutritional epidemiology. Secondly, notwithstanding any controversy, international guidelines now recognize the validity and endorse the use of these diets as a therapeutic nutritional approach, in appropriate patients. Thirdly, we have found that early de-prescription of diabetes medications is essential, in particular insulin, sulphonylureas, and sodium-glucose cotransporter (SGLT2) inhibitors. Fourthly, we encourage patients to eat ad libitum to satiety, rather than calorie counting per se. Furthermore, we monitor cardiovascular risk factors frequently, as with all patients with obesity or diabetes, but we do not necessarily consider an increase in low-density lipoprotein (LDL)-cholesterol as an absolute indication to stop these diets, as this is usually related to large LDL particles, which are not associated with increased cardiovascular risk. In the absence of large randomized controlled trials with cardiovascular and other hard endpoints, adopting a low-carbohydrate diet is a legitimate and potentially effective treatment option for patients with diabetes or obesity.