The model for tap-changing transformers currently in widespread use consists of an impedance, measured in a shortcircuit test with a nominal tap, in series with an ideal transformer. The use of this model in voltage stability studies leads to qualitatively incorrect results. For demonstration purpose, a small circuit and the concepts of maximum load, equivalent load impedance and voltage control effects are used. An improved model that takes into account laboratory results, circuit laws and voltage stability theory is proposed. Using IEEE test systems, different results are shown for the conventional and proposed models.