AbstractAimsAdopting a low‐ or very low‐carbohydrate (LCD or VLCD) diet in type 1 diabetes mellitus (T1D) is a controversial intervention. The main fear is that these diets may increase the risk of diabetic ketoacidosis. However, there is little data about the ketoacidosis risk and the level of physiological nutritional ketosis in individuals following these diets. We aimed to define the level of ketosis in those with T1D following carbohydrate restricted diets in a real‐world observational study.MethodsPatients with T1D who had self‐selected dietary carbohydrate restriction were enrolled from local clinics and were compared to those following an unrestricted regular carbohydrate control diet (RCCD). Participants completed a 3‐day diary, documenting food intake, ketones, and blood/interstitial glucose concentrations.ResultsParticipants were divided into three groups according to mean carbohydrate intake: VLCD (<50 g carbohydrates/day) n = 6, LCD (50–130 g carbohydrates/day) n = 6, and RCCD (>130 g carbohydrates/day) n = 3. Mean beta‐hydroxybutyrate (BOHB) concentrations were 1.2 mmol/l (SD 0.14), 0.3 mmol/l (SD 0.12) and 0.1mmol/l (SD 0.05) in the VLCD, LCD and RCCD groups, respectively (p = 0.02). Post hoc Dunn test demonstrated this reached statistical significance between the VLCD and RCCD groups (p = 0.02).ConclusionCarbohydrate restricted diets, in particular VLCDs, are associated with a higher BOHB level. However, the degree of ketosis seen is much lower than we expected, and significantly lower than the level typically associated with diabetic ketoacidosis. This may suggest the risk of ketoacidosis is lower than feared, although safety will need to be evaluated further in large scale randomised trials.