Aims and objective. We aimed to study the diagnostic accuracy of the temporal artery thermometer vs. rectal temperature in a large group of children with and without fever, aged 0-18 years. Background. Many have studied the diagnostic accuracy of the temporal artery thermometer in children compared with a reference method, with contradictory outcomes. No studies have been carried out in a large group of children of all ages. Design. Diagnostic accuracy/validation study. Method. Children (0-18 years) with fever (T > 38 center dot 0 degrees C) were recruited through the emergency department and children with normal temperatures through the day-care department of the Children's Hospital. All children routinely had rectal temperature recordings. Temporal artery temperature was recorded shortly after the rectal recording. The mean absolute difference in temperature, the level of agreement (intraclass correlation coefficient) and the sensitivity and specificity of detecting fever were calculated. Results. A total number of 198 children (121 boys) participated, with a mean age of 5 center dot 1 (SD 4 center dot 7) years. Of those children, 81 had fever according to the rectal recording. Mean difference between temporal artery temperature and rectal temperature was -0 center dot 11 (SD 0 center dot 63) degrees C, with an agreement of 0 center dot 812. The sensitivity and specificity of the temporal artery thermometer for detecting fever were 67 center dot 9 and 98 center dot 3%, respectively. Conclusions. The diagnostic accuracy of the temporal artery thermometer in detecting fever in children of all ages is low. Relevance to clinical practice. We do not recommend replacement of standard clinical thermometers with temporal artery thermometers.