It has been claimed that good environmental performance can improve firms' economic performance. However, because of e.g. data limitations, the methods applied in most previous quantitative empirical studies of the relationship between environmental and economic performance of firms suffer from several shortcomings. We discuss these shortcomings and conclude that previously applied methods are unsatisfactory as support for a conclusion that it pays for firms to be green. Then we illustrate the effects of these shortcomings by performing several regression analyses of the relationship between environmental and economic performance using a panel data set of Norwegian plants. A simple correlation analysis confirms the positive association between our measures of environmental and economic performance. The result prevails when we control for firm characteristics like e.g. size or sub-industry in a pooled regression. However, the result could still be biased by omitted unobserved variables like management or technology. When we control for unobserved plant specific characteristics in a panel regression, the effect is no longer statistically significant. Hence, greener plants perform economically better, but the analysis provides no support for the claim that it is because they are greener. These empirical findings further indicate that a conclusion that it pays to be green is premature.