Assessing the quality of existing industrial control loops, or comparing between two alternative controller designs is becoming an important routine auditing task for the control engineer. While most of the research and commercial activity in CPA has been applied to linear systems to date, those researchers investigating nonlinear systems fall into one of two groups. The first group focussed on the diagnosis of a common specific nonlinearity, namely valve stiction [1-3], while the second group tried to establish the minimum variance performance lower bound (MVPLB) [4-8]. In this paper we will propose a new CPA performance index for general nonlinear models based on an ANOVA-like variance decomposition method. The results of two simulation examples illustrate that the proposed methodology is efficient and accurate. (C) 2010 Elsevier Ltd. All rights reserved.