A recurrent concern raised by the European GHG Emissions Trading System (ETS) is the fear of EU industry competitiveness loss: a loss in domestic production and a loss in profits. This paper analyses how production and profits in the European cement industry may depend upon allocation approaches. We analyse two contrasting allocation methods of free allowances. Under "grandfathering", the number of allowances a firm gets is independent of its current behaviour. Under "output-based allocation", it is proportional to its current production level. Whereas almost all the quantitative assessments of the EU ETS assume grandfathering, the real allocation methods used by Member States, notably because of the updating every five years and of the special provision for new plants and plant closings, stand somewhere between these two polar cases. We study the impacts of these two polar allocation methods by linking a detailed trade model of homogeneous products with high transportation costs (GEO) with a bottom-up model of the cement industry (CEMSIM). The two allocation approaches have very different impacts on competitiveness and emissions abatements. Grandfathering 50% of past emissions to cement producers is enough to maintain aggregate profitability (EBITDA) at its business-as-usual level, but with significant production losses and CO2 leakage. For an output-based allocation over 75% of historic unitary (tCO2/tonne-cement) emissions, impact on production levels and EBITDA is insignificant, abatement in the EU is much lower but there is almost no leakage. Policy needs to recognise to what extent different allocation approaches may change the impacts of emissions trading, and adopt approaches accordingly.