Introduction: Among those prediction methods for estimating size of unerupted canine and premolars, Moyers and Tanaka-Johnston methods are most widely used, though the method-using radiograph is more accurate. Significant differences in tooth size exist among different ethnic groups. Therefore, an accurate prediction method for American white maybe less accurate for Mongolians. Objective: The purposes of the study were to develop new regression equations for predicting the size of the unerupted permanent canine and premolars in Mongolian children. Methods: Eighty-four males and eighty-three females in total 167 sets of dental cast of the permanent dentition aged 14.20.96 years were included in this study. The criteria of subjects selecting were following: all permanent teeth erupted (third molars may not be present), Angle I class fi rst molar relationship, no fi lling and no caries, no history of orthodontic treatment. The mesio-distal crown diameters were measured with a caliper to an accuracy of 0.05 mm. The data were processed with SPSS16.0 software package. Independent t-test and regression analysis were used for statistical analysis. Results: The development new linear regression equations with predictor sum 21|12 for predicting the size of the canine-premolar segment was based on the normative standard of mesio-distal crown diameters of permanent teeth in Mongolian children. The following new regression equations were developed for unerupted canines and premolars in Mongolian children Y=9.07+0.57*x in maxillary, Y=8.43+0.55*x in mandible arch. Discussion: This study confi rmed that the use of Moyers and Tanaka-Johnston prediction methods for mixed dentition analysis among Mongolian children were unsuitable. Both methods underestimated the size of caninepremolar segments. Conclusions: The following new regression equations were developed for interrupted canine and premolars in Mongolian children: y=9.11+0.57*x in maxillary and y=8.43+0.55*x in mandible arch.