The purpose of this article is to investigate, from the empirical side, whether the selection of a method of assessment of bank branches operational performance determines the results. For this purpose the theoretical model was constructed exposing the definition of operational efficiency and the approach to its measuring. The authors focus on three standard methods of efficiency analysis: Data Envelopment Analysis (DEA), Stochastic Frontiers (SF) and Corrected Least Squares Method (COLS). The empirical research was carried out with the use of the data for one of the largest Polish retail banks covering 37 bank branches during the first half of 2005. The selection of variables included number of operations, financial and non-financial inputs such as number of employees, space, number of ATMs, etc. The results generated by SF and COLS procedures were similar and differed significantly from DEA results. On the basis of structure analysis it appeared that DEA method tended to overvalue the efficiency of the small branches, which leads to statistically significant differences between the results.