This study aims at assessing the environmental performance the U.S. agriculture with respect to GHG emissions across states. To reach this objective, this paper utilizes alternative non-parametric approaches: a graph measure of technical efficiency under strong disposability and weak disposability and a modified output oriented Malmquist index. The graph measure of technical efficiency accounting for undesirable outputs reveals that regulations of agriculture GHG emission would be effective in all states but Delaware, as they would be binding and impose a ‘cost’ in terms of reduction of desirable output. Results show also that on average regulations would improve technical efficient for about 3.5%. States operating on the frontier shift from one to seven when the regulation is accounted for. But the opportunity cost of binding to this regulation amounts to 3.7% reduction of agricultural output. The approach Malmquist index and its components reveal that on average the efficiency change has been invariant to the treatment of the undesirable output as input. The average productivity growth is 2.2 percent when GHG emissions are treated as input whereas it is 2 percent when they are complementally ignored. In both cases, the productivity growth is driven by technological change.