In this study we apply Data Envelopment Analysis (DEA) to analyse efficiency in presence of both desired and undesired outputs. We use register data from the Swedish national environmental accounts. The study concerns the Swedish pulp and paper industry. To compute efficiency, we use a directional distance function approach that allows scaling for different directions, both desired and undesired output directions under the assumption of equal importance. The analysis show that efficiency have increased over time, both in terms of desired output and in terms of environmental efficiency. The first analysis measure inefficiency if scaling is done with equal importance given to desired output and undesired output. This analysis shows an inefficiency of around 12 % meaning that there is a 12% potential to simultaneously reduce pollution and increase production of pulp and paper. In the second analysis scaling is only allowed in the direction of undesired outputs. This analysis shows that the average potential reduction in pollution is 22%. If, on the other hand, the potential is computed in terms of only desired output keeping undesired output constant, the average potential increase in desired output is 14%. In a forth analysis we discuss the potential to evaluate the costs of environmental regulations using the DEA methodology.Keywords: Environmental efficiency, directional distance function,