The goal of this work is to improve understanding of wheel/rail lubricating mechanisms and propose an approach for use in the future to evaluate new lubricants suitable for the wheel flange/rail gauge corner contact. This work is carried out in the framework of the European research program 'Urban Track' (project no FP6-31312). Previous works on wheel-rail contacts have highlighted the presence of a natural third body composed of detached particles from both the wheel and the rail. In the case of the active (i.e. where wheel-rail contact occurs) rail gauge corner and the wheel flange contact, this third body is quickly mixed with the active flange lubricant, resulting in a mixture that can be considered as the 'real' lubricant of the contact. Tribological tests are performed on a reduced scale simulator in view to investigating the formation of such a mixture and its range of action. Its rheology 'in situ' is studied as is the detachment of particles (from rail and wheel specimens). The test conditions reproduced were close to those found in wheel-rail contacts, i.e. high shearing and pressure conditions. The migration and transport of the lubricant by the wheel specimen are highlighted during these tests and a 'pasty liquid' is quickly formed. It is assumed that this 'liquid' is composed of detached particles mixed with the initial lubricant leading to the formation of a mixture whose flows allow the initialization of a smooth surface. The latter facilitates the actions of oil additives, leading to a low friction coefficient. This sequence may produce an 'optimized' mixture formed in situ. This is coherent with analyses performed on site. Hence it is necessary to reconstitute a specific surface of the rail gauge corner, i.e. the right third body for lubrication.