This study was carried out to assess the accuracy of three devices namely, liquid crystal forehead, digital electronic axillary and infrared tympanic thermometer, using a glass-mercury rectal thermometer as the control. The subjects were two hundred children aged 0-48 months. The mean rectal temperature was 38.0 +/- 0.91 degrees C; forehead, 37.83 +/- 0.94 degrees C; tympanic, 37.77 +/- 0.95 degrees C, and axillary, 37.71 +/- 0.86 degrees C. Compared to the rectal temperature, all values were significantly lower (p < 0.05). Forehead, tympanic and axillary temperature differed from rectal temperature by at least 0.5 degrees C in 33.33 per cent, 23.5 per cent and 31.5 per cent of subjects, and at least 1 degrees C in 22 per cent, 1 per cent and 6 per cent of subjects respectively. Accuracy in detection of fever was 79 per cent for forehead, 85.5 per cent for tympanic and 84 per cent for axillary thermometry. Sensitivity of the three devices was 67-83 per cent in detection of fever and 64-77 per cent in detection of high fever. Tympanic thermometry had the best performance while forehead thermometry had the poorest. After using revised diagnostic threshold temperature by ROC curves, sensitivity of each device improved but accuracy was nearly the same. It is concluded that the three devices are not suitable as a substitute for a glass-mercury rectal thermometer in assessment of fever in infants and young children.