Background: Clinically apparent cerebral oedema during diabetic ketoacidosis (DKA) is rare and more common in children and young adults. Subclinical oedema with mild brain dysfunction is more frequent, with unknown long term effects. Rapid tonicity changes may be a factor although not well studied. Guidelines recommend capping hypertonicity resolution at ≤ 3 mOsmol/kg/h. Objectives: To audit current DKA management in the emergency department (ED) and in the intensive care unit (ICU) for tonicity benchmark compliance, and to determine interactions between plasma tonicity, plasma glucose concentrations and blood haemoglobin concentrations. Methods: Twenty-five adult DKA admissions from ED to ICU were studied retrospectively. Blood gas and electrolyte data were sequenced for 24 hours from first ED blood sample. Results: Sampling was frequent (median, 11 times per day; range, 6-26). Tonicity reduction was largely accomplished by the first ICU blood sample and exceeded 3 mOsmol/ kg/h in 72% of admissions. Correlation with haemoglobin reduction (haemodilution) rates exceeded correlation with glucose rates (R2 = 0.52 v 0.38). In benchmark noncompliant admissions, haemodilution was more rapid (6.1 g/L/h v 2.1 g/L/h; P = 0.001). Although also true of glucose reduction (4.5 mmol/L/h v 2.2 mmol/L/h; P = 0.007), there was no interaction between haemodilution and glucose reduction (R2 = 0.09). Conclusions: Major tonicity reductions often exceeding guidelines were common by ICU admission. Correcting DKA-induced hypertonicity at ≤ 3 mOsmol/kg/h requires controlled hyperglycaemia correction and, based on our data, avoidance of high fluid replacement rates; for example, sufficient to reduce haemoglobin concentrations by > 3 g/L/h, unless there is evidence of intravascular hypovolaemia.