This paper aims to estimate the eco-efficiency of agriculture in the state of Goias, Brazil, using the Data Envelopment Analysis (DEA) method, combined with Directional Distance Function (DDF). Therefore, data obtained from the Agricultural Census of 2006 of all 246 municipalities in the state of Goias was used, with eight variables: four inputs, three desirable outputs, and one undesirable output. The results allowed us to map and classify the municipalities into four categories: i) eco-efficient; ii) low ecoinefficiency; iii) average eco-inefficiency; and iv) high eco-inefficiency. The results showed that sixty-eight municipalities were considered eco-efficient, with Alto Paraíso de Goias and Aparecida do Rio Doce appearing as benchmarks for half of the sample cities. The municipality with the worst performance was Quirinopolis. The overall score of eco-inefficiency in the State was of 0.77, which indicates that resource savings with labor, capital and other inputs could reach R$ 1,290,264,669.94 and 194,047 ha of land. Concurrently, it is possibleto increase revenue by R$ 2,835,872,090.34, and 1,534,833 ha of protected area, while reducing degraded areas by 20,470 ha. Thus, these results support the hypothesis that the apparent antagonism between economic and environmental objectives does not prevent the formulation of policies consistent with the improvement of environmental economic sustainability.