To better understand the role and reliability of axillary temperature measurements in clinical real life, axillary and rectal measurements in infants presenting in a private pediatric practice because of fever were compared. Prospectively, 169 infants (81 girls), median 9 (interquartile range 6-13) months of age, were examined at room temperature (20-24 °C). Two left and two right axillary, as well as two rectal measurements were taken with a digital thermometer and subsequently averaged. The median and interquartile range for axillary and rectal measurements were 36.9 (36.3-37.6) °C and 38.2 (37.4-38.9) °C, respectively (p < 0.0001). The limits of agreement in the Bland-Altman plots were 0.32 to 1.98 °C, with a mean bias of 1.15 °C. Axillary thermometers showed a good sensitivity for detecting rectal temperature > 38 °C (95%) but limited specificity (75%), with an area-under-the-curve of 0.95.Conclusions: Axillary readings are always lower than rectal ones, the limits of agreement are quite wide. Axillary readings can be used for screening but critical measurements should be confirmed by more reliable methods. What is Known • In infants and toddlers, temperature has been traditionally taken rectally. • Axillary measurements are better accepted and are recommended in current guidelines. What is New • Axillary temperature was always lower than rectal temperature. • The limits of agreement of axillary thermometers are wide. • Axillary thermometers have a good sensitivity but limited specificity and are therefore adequate for fever screening.