Several index decomposition methods are commonly employed to provide a top-down view of energy consumption trends in manufacturing industry. These approaches typically use value added data for an industrial sector to decompose energy trends into structural, intensity and activity effects. Additionally in Europe a commonly employed top-down indicator called ODEX uses units of physical output, rather than value added, to analyse energy efficiency developments only. Therefore it has been difficult to compare ODEX directly to decomposition approaches. This paper presents a new decomposition method called VALDEX, based on the existing ODEX methodology, but using value added data. Extending ODEX to a full decomposition method allows tests commonly used in index decomposition theory to be applied and enables direct comparison with other methods. This helps evaluate the robustness of the existing ODEX methodology. Using industry data from three European countries, the results yielded by five decomposition methods are compared. In the cases examined, both the Laspeyres and VALDEX methods have significant residuals. Laspeyres consistently overestimates total energy consumption while VALDEX underestimates it. Methods that produce small or no unexplained residuals give converging results for each effect for the countries analysed, and provide a more reliable view of energy trends. (C) 2010 Elsevier Ltd. All rights reserved.