The rectal temperature is frequently measured in the morning on a daily basis in hospitalized dogs. Besides comparing the measured temperature to a population-based reference interval, it may also be of interest to evaluate the differences between consecutive measurements in order to follow the health status of a particular dog. In veterinary clinical chemistry, a parameter referred to as the critical difference, which can be used to judge if the difference between two consecutive measurements may be safely ascribed to random variation or not, has recently been introduced. The purpose of the present study was to obtain an estimate of the critical difference of the rectal temperature based on measurements of the rectal temperature in clinically healthy dogs in a hospital environment. The rectal temperature was measured in a standardized manner on 6 different days in 7 dogs. The total variance was estimated as 0.1015. The inter-individual variance was 0.0428, and the intra-analytical variance including the analytical variance was 0.0587. From these components of variance, the critical difference was calculated as 0.69 degrees C. Thus, the rectal temperature measured in the morning in hospitalized dogs can be expected to fluctuate about 0.7 degrees C from day to day due to random variation alone.