OBJECTIVE To compare the results of reading body temperatures with a tympanic infrared thermometer and a rectal mercury thermometer in children in an emergency department. METHODS Prospective comparative study. METHODS St. Elisabeth Hospital, Tilburg, the Netherlands. METHODS In children up to 11 years of age seen in the emergency room between 1 January 1994 and 1 April 1994, the body temperature was measured with a rectal mercury thermometer as well as with a tympanic infrared thermometer. Data were collected on temperature read, clinical picture on arrival (not ill, ill, seriously ill) and appearance of the tympanic membrane (signs of acute otitis media, presence of cerumen). For the statistical comparison, the differences between the findings of the two methods were plotted against the means. The sensitivity and specificity of the results of tympanic measurement in relation to the values read rectally were determined. RESULTS Data were collected on 213 children, of whom 19 were younger than 3 months, 46 between 3 and 12 months, and 148 between 1 and 11 years. The mean temperatures measured with the rectal and tympanic thermometers were 38.01 and 38.03 degrees C, respectively. The mean difference between the rectal and tympanic temperatures was -0.013 degree C. The correlation between the rectal and tympanic temperatures was high (r = 0.86; P = 0.0001). The results were the same in groups differing in age, severity of disease and appearance of the tympanic membrane. The sensitivity of the tympanic measurement for fever (rectal temperature > 38.0 degrees C) was 80.6% with a specificity of 92.5%. The sensitivity was 83.8% when a rectal temperature > 38.5 degrees C was taken as the criterion, with a specificity of 95.9%. CONCLUSIONS The tympanic infrared measurement in children in an emergency department gave the same results as rectal measurement using a mercury thermometer.