The European Renewable Energy Directive (EU RED) requires biofuels to reduce greenhouse gas emissions (GHG) by 35% compared to fossil fuels in order to count towards mandatory biofuel quota or to be eligible for financial support schemes. This reduction target will rise to 50% in 2017. For biofuel producers this implies that they want or need to calculate their emissions. The purpose of this paper is to compare two calculation tools for economic operators that are on their way to the market: the "BioGrace tool" and the "Roundtable on Sustainable Biofuels (RSB) GHG tool" for GHG calculations under the Renewable Energy Directive (both of which are freely available). Greenhouse gas emissions from four production pathways were calculated: ethanol from wheat, ethanol from sugarcane, biodiesel from rapeseed and biodiesel from palm oil. In addition, three land use change (LUC) scenarios were calculated: for expansion of the biofuel cultivation area to grassland and to forest (10-30% canopy cover) and for improvement of agricultural practices. Both tools follow the methodology of the European Renewable Energy Directive and exactly the same input data along the production chain was used. Despite this, the results were significantly different. GHG emissions of the pathway ethanol from wheat were 21% lower when calculated with the BioGrace tool than with the RSB GHG tool. Differences were most pronounced in the cultivation phase with 20% deviation between the tools for biodiesel from palm oil and 35% deviation for ethanol from wheat and sugarcane. In practice this means that an economic operator can enhance the GHG performance of his biofuel by 20-35% by using a different calculation tool without improving the production process. We identified the use of different standard values in the two tools, in particular for the production of N-fertilisers, for chemicals and electricity and one methodological choice regarding the calculation of field N2O emissions as source of these differences. This methodological point is not specified in the Renewable Energy Directive, giving economic operators and tool developers free choice. GHG emissions from land use changes varied by -14% to 49% due to differences in carbon stock data, methodological differences in allocation and a lack of precise land use type definitions. We conclude from the results that there is a need for a deep harmonisation in the calculation process that goes beyond the methodological framework set up in current legislation. These findings are relevant because they show a policy gap, a regulatory gap that needs to be addressed by policy makers in order to guarantee a level playing field on the market and to create an incentive to improve the GHG performance of biofuel production. (C) 2012 Elsevier Ltd. All rights reserved.