OBJECTIVE To assess the reproducibility of Tronzo and AO/ASIF classifications for transtrochanteric fractures, in order to determine the most appropriate classification for clinical application, and to evaluate the influence of the level of experience of the observers in the agreement between evaluations. METHODS We selected 30 radiographic images of transtrochanteric fractures of the femur, which were presented to two groups of observers, one formed by expert physicians and the other by resident physicians. RESULTS When evaluated together, Tronzo classification obtained a Kappa value of 0.44. The same classification assessed by the expert group obtained a value of 0.46, while the group of residents' value was 0.44. Evaluating the AO/ASIF classification of the complete pool analysis the value found was 0.42. For the same classification, analyzed by the expert group, obtained a value of 0.41, and by the group of residents, the Kappa value achieved was 0.42. However, when analyzed in its simplified form, the AO/ASIF classification obtained Kappa values of 0.70 (pooled analysis), 0.68 (experts) and 0.72 (residents), considered concurrent. CONCLUSIONS The AO/ASIF simplified classification showed substantial reproducibility and is, therefore, recommended as the most suitable for clinical application. The level of experience of the observers did not influence significantly the agreement between evaluations. Level of Evidence III, Diagnostic Study - Investigating a Diagnostic Test.