Intravenous insulin protocols are increasingly common in the intensive care unit to maintain normoglycemia. Little is known about the accuracy of point-of-care glucometers for measuring glucose in this patient population or the impact of sample source (capillary, arterial, or venous whole blood) on the accuracy of glucometer results. We compared capillary, arterial, and venous whole blood glucose values with laboratory plasma glucose values in 20 patients after cardiac surgery. All 4 samples (capillary, arterial, and venous whole blood and laboratory plasma glucose) were analyzed hourly for the first 5 hours during intravenous insulin therapy in the intensive care unit. There were no significant differences between median capillary whole blood (149 mg/dL [8.3 mmol/L]) and laboratory plasma (151 mg/dL [8.4 mmol/L]) glucose levels. The median arterial (161 mg/dL [8.9 mmol/L]) and venous (162 mg/dL [9.0 mmol/L]) whole blood glucose levels were significantly higher than the median laboratory plasma glucose level. Capillary whole blood glucose levels correlate most closely with laboratory plasma glucose levels in patients receiving intensive intravenous insulin therapy after cardiac surgery.