In this paper we adopt a new technique to evaluate economic efficiencies of Taiwan’s banking sector, proposed by Fried et al. (2002). The technique involves a three-stage analysis and is able to adjust the effects of environmental variables and statistical noise on banks’ performance under the framework of data envelopment analysis (DEA). The first and the last stages involve the conventional DEA, where the former uses the original inputs and outputs while the latter uses the adjusted inputs, which account for the impact of the environmental effects and the statistical noise found in the second stage, and unadjusted outputs to reevaluate producer performance. The second stage plays a pivotal role, during which the variation in performance is decomposed into three components, i.e., the environmental effects, managerial inefficiency, and the statistical noise.