The present study is concerned with the critical difference, which may help to judge whether or not the difference between two consecutive measurements with a certain probability (i.e. 95 percent) may be ascribed to natural variation. Knowledge of the applicability of the critical difference in veterinary medicine is sparse and therefore, to justify future use of the critical difference, it is important to know whether or not the critical difference performs as expected. The hypothesis to be tested in this study was that at least 95 percent of the differences between consecutive measurements, that have been obtained in animals where the component measured is known to be unchanged, should be within the critical difference. From previous studies it was known that a low‐sodium diet had no influence on the plasma potassium concentration. The critical difference of the plasma potassium concentration in dogs was calculated as 0.5 mmol/l. using weekly measurements of this plasma component in a group of twenty healthy dogs. To test the hypothesis, this value was compared to the differences between consecutive weekly measurements of this plasma component in another group of eight dogs fed a low‐sodium diet for five weeks. In agreement with previous studies, the plasma potassium concentration in the eight dogs did not change significantly during the feeding experiment. Of the fourty differences between consecutive weekly measurements, thirty‐six were within the critical difference. This number was not different from the number expected from the hypothesis and thus, the critical difference performed as expected. Furthermore, the present study also indicated that a critical difference calculated from one group of dogs can be used to evaluate differences between consecutive measurements in another group of dogs under other dietary and management conditions.