We measured the skin temperature over the forehead, jugular notch, and neck and compared all three infrared skin temperature sites, axillary digital, and infrared tympanic readings with the rectal temperature in children in a pediatric emergency observation unit. This study enrolled 139 patients ranging in age from 1 month to 4 years seen in the pediatric emergency observation room. The temperatures in the rectum and axilla were measured with a digital thermometer, and a non-contact infrared thermometer was used to measure the temperature of the skin over the lateral side of neck, jugular notch, forehead and the tympanic temperature. During the examination of each child, 17 temperatures were measured within 10 minutes, including six (three for each ear) tympanic and three infrared skin temperatures at each of the three sites, one axilla, and one rectal reading. Fever was defined as a rectal temperature >= 38 degrees C, and 56 patients fulfilled this definition. All measurements were compared with the rectal measurement using Bland-Altman plot analysis. The ability of each method to predict a rectal fever was calculated using receiver operating curve (ROC) analysis. The rectal readings were slightly and significantly (p<0.01) higher than all other measurements overall. The Bland-Altman plots revealed correlations between the rectal temperatures and the axillary, left and right tympanic, and all infrared skin temperatures, but the latter measurements were not in good agreement with standard readings. The highest areas under the curve (AUCs) were 0.942, 0.882, 0.857, and 0.806 for the axillary, right and left tympanic and infrared-neck thermometers, respectively. Tympanic thermometry is a practical method for measuring temperature in children 1 month to 4 years of age in a busy emergency observation room instead of axillary thermometry, which is more accurate but time-consuming.