Two in situ concrete/claystone interfaces were sampled at the laboratory level in the Andra Meuse/Haute Marne (France) Underground Research Laboratory (URL) in order to study five years of interactions between Callovian-Oxfordian (COx) claystone and two cementitious materials (concrete bottom slab and shotcrete on the walls of the main gallery), with a specific focus on sulfur. Combined mineralogical, chemical and sulfur isotopic investigations were carried out to define the degree of the perturbation of the sulfur system in the claystone and in both the cementitious materials. At both interfaces, results show that the main perturbation on the claystone side is the formation of scarce μm-sized gypsum, the sulfur content of which is essentially derived from pyrite oxidation. The distribution of gypsum is highly correlated with the fissure network of the damaged zone due to excavation of the gallery. Its presence is also often associated with a loss of cohesion of the concrete/claystone interface. Due to the small amounts of gypsum and its μm-size, measurements were performed by ion microprobe. Adaptations were needed on account of the reactivity of gypsum and sulfates in general under the beam. The use of ion microprobe analysis provided evidence of high local isotopic heterogeneity that could be attributed to kinetic fractionation effects. Some analyses suggest a minor contribution of dissolved sulfates in pore water of claystone and possibly of concrete. The perturbation on the concrete side is marked by a significant increase in the bulk sulfur content within three millimeters of the interface with the claystone, showing a sulfur gradient from claystone to concrete. The main objective of this work was to define the extent of the chemical and mineralogical perturbations, taking into account in situ URL conditions, i.e. hydrodynamic conditions (shotcrete sprayed on the gallery walls and subjected to ventilation of the galleries), damaged zone of claystone induced by excavation and multi-scale heterogeneity. This identifies the relevant parameters to better constrain predictive reactive transport models of concrete/claystone interactions at a realistic scale.