BACKGROUND: Despite increasing use of tympanic thermometers in critically ill patients who do not have a pulmonary artery catheter in place, variations in measurements obtained with the thermometers are still a problem. OBJECTIVE: To compare the range of variability between tympanic and oral electronic thermometers. METHODS: Subjects were a convenience sample of 72 patients admitted to a 24-bed adult medical-surgical intensive care unit. For each patient, temperatures were measured concurrently (within a 1-minute period) with an oral (Sure Temp 678) thermometer, a pulmonary artery catheter (Baxter VIP Swan-Ganz Catheter), and 2 tympanic (FirstTemp Genius II and ThermoScan Ear Pro-1) thermometers. Each subject was used up to 3 times for data collection. Measurements obtained with the oral and tympanic thermometers were compared with those obtained with the pulmonary artery catheter. Nonparametric analysis of data was used. RESULTS: The magnitude of error for the ThermoScan tympanic thermometer differed significantly from that of the Genius II tympanic thermometer and the SureTemp oral thermometer (P < .001). Application of the Bland and Altman method to frame the data on the basis of an accuracy tolerance zone of +/-0.5 degrees C indicated variability with both the oral and tympanic methods. The overall degree of variability was lower for the oral thermometer. CONCLUSIONS: Oral thermometers provide less variable measurements than do tympanic thermometers. Use of oral thermometry is recommended as the best practice method for temperature evaluation in critical care patients when measurement of core temperature via a pulmonary artery catheter is not possible.