AbstractAimsThe study examined the prevalence and degree of lactate elevation in diabetic ketoacidosis, and explored which biochemical abnormalities predicted L‐lactate levels.MethodsWe reviewed episodes of diabetic ketoacidosis from 79 diabetes patients (one episode per patient). Separate univariate linear regression models were specified to predict lactate level from each of nine biochemical variables. Significant predictors from the univariate models were included in a final multivariate linear regression model to predict lactate levels.ResultsMean (SD) lactate level was 3.05 (1.66) mmol/L; about 65% of patients had lactate levels >2 mmol/L. In the final multivariate linear regression model (R2 = 0.45), higher lactate levels were associated with greater hydrogen ion concentration (standardised β = .60, t = 4.16, p < 0.0001), higher blood glucose (standardised β = .28, t = 2.67, p = 0.009) and lower glomerular filtration rate estimated from creatinine (standardised β = −.23, t = 2.29, p = 0.025). Bicarbonate, beta‐hydroxybutyrate, body mass index, mean arterial pressure and calculated osmolality were not significant predictors of lactate level. There were three distinct patterns of lactate levels with treatment of diabetic ketoacidosis: group 1 = gradual decline, group 2 = initial increase and then decline and group 3 = initial decline followed by a transient peak and subsequent decline.ConclusionsElevated lactate level is the norm in patients with diabetic ketoacidosis. Higher blood glucose levels and higher hydrogen ion concentrations are related to greater lactate. With treatment, there are different patterns of decline in lactate levels.