Worldwide, a large number of contaminated site are reported due to anthropogenic activities and the intense use of chemicals. These sites need an appropriate management and remediation options that take in consideration, not only environmental risks, but also the linked socio-economic aspects. In the last years, phytoremediation techniques have been studied and selected as suitable and environmentally-friendly method to restore contaminated ecosystems. The selection of plants for phytoremediation is mostly done through ex situ pots experiments in greenhouse, growth chambers or in hydroponic conditions. Moreover, pots experiments are usually short-term study, while in situ experiments can last several years or even decades. For these reasons, the aim of this work was to compare plant response to heavy metal stress using plants grown directly on the contaminated fields and plants grown in pots containing soil from the same contaminated site. Among different species of studied plants, Miscanthus x giganteus seems to be a promising crop thanks to its demonstrated capacity to combine both environmental and economic benefits. In this experiment plants have grown in the Metaleurop Nord site, in Northern France, which has been suffering for more than a century for atmospheric emissions from two smelters present in the area. In this site, soils present a concentration of heavy metal which is 20 to 50 times higher than the regional background. In parallel to the in situ experiments, a two years ex situ experiment have been set up. Miscanthus plants were grown in pots containing contaminated soils from the Metaleurop site. The purposes were to evaluate (i) heavy metal uptake, (ii) translocation to different plant organs, (iii) and the metal impacts on plant health, using a set of biomarkers (evaluation of oxidative stress, DNA degradation, alteration of photosynthetic pigments, and production of secondary metabolites…). Results showed significant differences in almost each biomarker considered, demonstrating lower metal uptake, better adaptation, and a lower stress level in plants grown in field. Moreover, it is important to highlight that after several months, plants in pots demonstrated stress even in control samples compared to the in situ plants. This suggests that pot experiments do not reveal what happens on the field, by overestimating plant metal uptake capacity but also by modifying plant response to metal stress. This work demonstrates the importance of leading in situ experiment to understand fully how plants behave and adapt to site conditions, and to select better plants suitable for phytomanagement. More generally, it raises the question of the relevance of selecting plants with pots experiments.