To study the reliability of the Tanaka-Johnston mixed-dentition analysis in Turkish individuals and develop a new population-adjusted regression equation for predicting the size of unerupted canines and premolars. The new regression equations were tested to establish their validity in Turkish children. The mesiodistal dimensions of teeth on study casts of 12- to 14-year-old Turkish children were measured. The actual tooth measurements were then compared with predicted values using the Tanaka-Johnston equation. Subsequently, the new regression equations were tested in a group of 50 individuals to determine their validity. The Tanaka-Johnston regression equations overestimated the actual mesiodistal widths of permanent canines and premolars. With the new regression equations, no statistically significant differences were observed between actual and predicted widths. The Tanaka-Johnston prediction method is not accurate when applied to a Turkish population. The new equations are reliable for predicting the width of unerupted canines and premolars in Turkish individuals. © 2009 Quintessence Publishing Co, Inc. All Rights Reserved