The reduction in the number of distractors may be a useful tool to improve the psychometric quality of a test or to shorten it without affecting content validity. The purpose of this study is to put forward criteria for the reduction of distractors as suggested in the characteristic curves analysis of an Item Response Theory model: Thissen and Steinberg's Multiple Choice (1997). Possible unsuitable distractor behaviors were classified in two groups: growing curve and horizontal curve. These categories allow to make a decision on which alternative to remove based on the degree of unsuitability of the distractor. This procedure leads to a more homogeneous whole with more plausible options. It is discussed which advantages there are in the implementation of these guidelines at different moments in the test construction or adaptation.