Loreau & Hector, (2001) proposed a framework to estimate the net biodiversity effect (over-or under-yielding) on the productivity of plant communities and to further decompose this effect into dominance and complementarity effects. Here we applied this method to analyse the performance of eight cereal-legume intercropping experiments grown in diverse environmental conditions in France, with three combinations of species mixtures: i) durum wheat/faba bean ii) soft wheat/pea and iii) durum wheat/pea. For each species mixture, we tested whether N fertilisation influenced dominance or complementarity effects using a Bayesian approach. Our results showed that the Biodiversity Effect (BE) was positive in the three intercrops under unfertilised conditions, mainly thanks to complementarity effect. The relative importance of BE and both of its components depends on the species intercropped. Finally, we showed that the impact of N fertilisation on BE of intercrops strongly depends on the competitive ability of the species intercropped.