This paper measures resource use efficiency of electricity generating plants in the United States under the SO2 trading regime. Resource use efficiency is defined as the product of technical efficiency and environmental efficiency, where the latter is the ratio of good output (electricity) to bad output (SO2) with reference to the best practice firm, i.e., one that is producing an optimal mix of good and bad outputs. This concept of environmental efficiency is similar to that of output oriented allocative efficiency. Using output distance functions we compare three methods for the calculation of resource use efficiency, namely, stochastic frontier analysis (SFA), deterministic parametric programming and nonparametric linear programming. This paper reveals the strengths and weaknesses of these methods for estimating efficiency. Both SFA and linear programming approaches can estimate the efficiency scores. For plants in the dataset the overall geometric mean of the three methods for technical efficiency, environmental efficiency and resource use efficiency is 0.737, 0.335 and 0.248, respectively. The rank correlation coefficient between technical efficiency, environmental efficiency and resource use efficiency is 0.213, 0.617 and 0.877, respectively. The regression analyses of performance across plants shows units in phase I of the SO2 trading programme are negatively related to measures of economic and environmental performance. This suggests that the market for SO2 allowances, per se, may not be minimizing compliance cost. We also find that a decrease in SO2 emission rates not only increases environmental efficiency but also leads to an increase in resource use efficiency. This finding concurs with the hypothesis that enhancement in the environmental performance of a firm leads to an increase in its overall efficiency of resource use as well.