This paper provides an empirical analysis of the determinants of energy efficiency in 32 power electric generation management companies over the period 2005-2009. The study uses non-parametric Data Envelopment Analysis (DEA) to estimate the relative technical efficiency and productivity change of these companies. In order to verify the stability of our DEA model and the importance of each input variable, a stability test is also conducted. The results of the study indicate that average technical efficiency of companies decreased during the study period. Nearly half of the companies (14) are below this average level of 88.7% for five years. Moreover, it is shown that the low increase of productivity changes is more related to low efficiency rather than technology changes. (C) 2011 Elsevier Ltd. All rights reserved.