Accurate measurement of body temperature is an important indicator of the status of critically ill patients and is therefore essential. While axillary temperature is not considered accurate, it is still the conventional method of measurement in Asian intensive care units. There is uncertainty about the accuracy of thermometers for the critically ill. We compared the accuracy and precision of bladder, axillary and tympanic temperature measurements in critically ill patients. A total of 73 critically ill patients admitted to the intensive care unit of a teaching hospital were prospectively enrolled. Every four hours, we measured body temperature at three sites (bladder, axillary and tympanic). If the patient had received an indwelling pulmonary artery catheter, blood temperature was also recorded and this was compared with bladder, axillary and tympanic temperature readings. For all patients, axillary and tympanic temperature readings were compared with bladder temperature readings. Accuracy and precision were analysed using Bland-Altman analysis. When blood temperature data was available, the mean difference between blood and bladder temperature readings was small (0.02±0.21°C). Compared with bladder temperature, mean difference for axillary temperature was -0.33±0.55°C and for tympanic temperature it was -0.51±1.02°C. For critically ill patients, recorded axillary temperature was closer to bladder temperature than tympanic temperature.