Functional-structural models of plant growth (FSPM) aim at describing the structural development of individual plants combined with their eco-physiological functioning (photosynthesis, biomass allocation, in interaction with the environment). The multi-biophysical processes described in FSPMs and their complex interactions make it difficult to identify the key processes, control variables and parameters. The objective of this study is to explore how global sensitivity analysis (SA) can help the design of such complex models in two aspects: first, quite classically, in the parameterization process and secondly by providing new biological insights and diagnosis. We consider a complex functional-structural model, NEMA(Bertheloot et al., 2009),describing Carbon (C) and nitrogen (N) acquisition by a wheat plant as well as C and N distributions between plant organs after flowering. This model has the specificity to integrate physiological processes governing N economy within plants: root N uptake is modeled following: High Affinity Transport System (HATS) and Low Affinity Transport System (LATS), and N is distributed between plant organs according to the turnover of the proteins associated to the photosynthetic apparatus. C assimilation is predicted from the N content of each photosynthetic organ. Inputs of Nitrogen fertilizers are fundamental to get high-yielding crops and a production of high quality with the required protein content. This required a proper understanding of root N uptake regulation and of N determinism on yield and production. Complex interactions exist between root N uptake, N remobilization to grains, and photosynthesis, whose regulatory mechanisms remain far from clear. In our application, analyses are conducted using Sobol's method and an efficient computation technique derived from (Saltelli, 2002), and several outputs of interest are considered. Moreover, since we consider a dynamic system, the evolution of the sensitivity indices is computed. The methodology developed is inspired by(Ruget et al., 2002)and first implies a module by module analysis. Basic biological modules are identified firstly. The full model involves around 80 parameters, and each module between 12 to 25 parameters. For each module and each output, the most important parameters (with the highest first-order Sobol indices) are identified. The interactions between parameters within the module also need to be identified. At this step, the least important parameters are then fixed in each module. The second step compares sensitivity produced by each module on the overall model outputs. The parameters for the SA are the ones selected as the most important from each module in the first step. The variance decomposition given by Sobol's method allows: ranking the effects of each module, but also the level of interactions between modules, regarding the output of interest. The consequences of this study are crucial in several aspects: for parameterization, stressing on which module and within each module on which parameter more care should be taken, but also on whether each module can be parameterized independently (from different experiments for example). Moreover, studying carefully the interactions between parameters and modules, dynamically, may reveal some biological phenomena of interests, non visible through simple simulations. In this regards, SA offers new tools in integrative biology.