This study evaluated the performance of a group of twenty-seven livestock production enterprises or decision making units (DMU) from fourteen CREA groups located in two contrasting areas of the country, with respect to their input cost efficiency. The study considered a production function with one product (beef) and three input variables (pastures, supplements and health costs). A data envelopment analysis (DEA) was applied the DMUs using average physical and economic information from three years. The DEA is a linear programming technique that enables to identify which DMUs, within a group, are more efficient, as they locate at the empirically estimated production frontier. The results showed that only six DMUs exhibited efficient input/output combinations (100%), three reached a 98% efficient, and one showed an efficiency rate of 92%. The inefficiencies were referred to the spending on supplementation or health. On the lower level, seven DMU showed efficiency levels lower than 60%, with one of them exhibiting an efficiency rate of less than 40%. The absence of linear correlations between the efficiency ratios and the descriptive variables characterizing DMUs, suggest that the differences were due to management factors that are related, in turn, to socio-economic characteristics of producers.