The amount of the preserved lateral pillar height (LPH) for assigning Herring grades is estimated during daily practice in Perthes disease. Despite this estimation, Herring classification has been reported to have a good interobserver agreement. The purpose of this study was to investigate whether the amount of preserved LPH has an effect on interobserver agreement. The results of surgeons' estimates were compared with objective measurement results. Good agreement was found among three experienced pediatric orthopaedic surgeons in 50 patients (1 vs. 2, kappa = 0.527; 2 vs. 3, kappa = 0.526; 1 vs. 3, kappa = 0.539). Twenty of these cases had a ratio of the LPH scattered between 0.45 and 0.60 and 0.90 and 0.99 (transition between group B and C, and group A and B). When the borderline cases were evaluated separately, the interobserver reliability was poor (1 vs. 2, kappa = 0.194; 2 vs. 3, kappa = 0.256; 1 vs. 3, kappa = 0.154), which may be explained by misperception of the preserved LPH. If the Herring classification is to be used as the prognostic indicator for deciding among the treatment alternatives, and LPH is the major determinant of the management, measurements should be used instead of estimates, especially in borderline cases.