OBJECTIVE To evaluate, in the initial and final stages of corrective orthodontic treatment, the predictability of the ANB, 1-NB, PNB and 1-NA during case individualization, which considers the characteristics of the patient, professional experience and the mechanics to be used. METHODS Ninety patients were selected at the State University of Londrina (UEL, Brazil), presenting Angle Class I and II malocclusions, treated with and without extraction of four premolars and divided into three groups: Horizontal, balanced and vertical. The cephalometric variables were evaluated in the initial, prognosis and final stages of treatment in order to observe the behavior of the estimates, or how they were higher or lower than the values obtained. RESULTS It was noticed the influence of the facial pattern on the behavior of the measures examined, the values proposed for the ANB were statistically different from values obtained at the end of the treatment; in the vertical group the final value was the one that most approached the proposed value; regarding 1-NB the values proposed with the Steiner analysis for the balanced and vertical groups were not achieved. For P-NB, there was no difference between genders. For 1-NA it was observed that the values obtained at the end of treatment differ from estimates in the three groups. CONCLUSIONS The limitations of the estimates of the measures do not invalidate its clinical or teaching use, if aware of its deficiencies, the analysis can be used with restrictions.