Crop growth models developed at field level are increasingly used in larger scale studies often in combination with other models to explore management options at the whole-farm scale or for integrated assessment at regional level (e.g. Giller et al., 2006; van Ittersum et al., 2008). Unclear is whether the mechanistic detail of dynamic simulation crop models is required for such applications or whether summary models may be sufficient. To address this issue and to identify relationships that need specific attention in future research, we compared simulations of crop production by two models with a different degree of mechanistic detail. (Résumé d'auteur)