Blood gas analysers are point-of-care testing devices used in the management of critically ill patients. Controversy remains over the agreement between the results obtained from blood gas analysers and laboratory auto-analysers for haematological and biochemistry parameters. We conducted a prospective analytical observational study in five intensive care units in Western Australia, in patients who had a full blood count (FBC), urea, electrolytes and creatinine (UEC), and a blood gas performed within 1 h of each other during the first 24 h of their intensive care unit admission. The main outcome measure was to determine the agreement in haemoglobin, sodium, and potassium results between laboratory haematology and biochemistry auto-analysers and blood gas analysers. A total of 219 paired tests were available for haemoglobin and sodium, and 215 for potassium. There was no statistically significant difference between the results of the blood gas and laboratory auto-analysers for haemoglobin (mean difference –0.35 g/L, 95% confidence interval (CI) –1.20 to 0.51, P = 0.425). Although the mean differences between the two methods were statistically significant for sodium (mean difference 1.49 mmol/L, 95% CI 1.23–1.76, P < 0.0001) and potassium (mean difference 0.19 mmol/L, 95% CI 0.15–0.24, P < 0.0001), the mean biases on the Bland–Altman plots were small and independent of the magnitude of the measurements. The two methods of measurement for haemoglobin, sodium and potassium agreed with each other under most clinical situations when their values were within or close to normal range suggesting that routine concurrent blood gas and formal laboratory testing for haemoglobin, sodium and potassium concentrations in the intensive care unit is unwarranted.