Over the last decade, efforts have been carried on to develop and evaluate versions of global terrestrial ecosystem models (GTEM) in which crop specificities are represented. The goal of this study is to evaluate the ability of the ORCHIDEE-STICS (Organising Carbon and Hydrology In Dynamic EcosystEms-Simulateur mulTIdisciplinaire pour les Cultures Standard) GTEM in simulating the observed seasonal variations and annual budgets of net ecosystem exchange (NEE), gross primary production (GPP) and total ecosystem respiration (TER) fluxes over seven wheat sites spanning a large climate gradient in Europe. Overall, the seasonal variations of GPP are well represented by the model, with 5 sites out of 7 exhibiting a correlation coefficient (R) value higher than 0.9 and a normalized standard deviation (NSTD) between 0.8 and 1.2. In comparison, the model performances for catching the seasonal variations of TER are lower, especially in terms of NSTD. Regarding the annual budgets, mean simulated deviations averaged over all sites do not exceed 10% and 15% of the observed annual mean budget, for GPP and TER, respectively. For NEE, the model capacity at estimating annual budgets is low, its mean deviation corresponding to ~35% of the observed mean value. This clearly shows that more accurate model estimates of GPP and especially TER are required for estimating NEE annual budgets. In this respect, past land-use and land-management changes are probably the most crucial processes to add, for getting soil carbon disequilibrium and more accurate NEE annual budgets. From a sensitivity analysis of the modelled fluxes to three management practices (plant variety, sowing date and fertilization intensity), we found that the fertilization is the most sensitive practice impacting the model performances of any flux, both in terms of seasonal variations and annual budgets.