The dynamics of a rail vehicle is driven by the interaction between the wheel and rail. Any change to, for example, the shape of the wheel-rail profile or the contact adhesion conditions will change the response of the vehicle. The condition monitoring challenge is to interpret these changes into useful condition information. This paper presents the ongoing research into model-based condition monitoring at the wheel-rail interface applied to two applications: (i) wheel-rail profile estimation; and (ii) low adhesion detection. The wheel-rail profile estimation was carried out on a linearised simulation model that included a nonlinear conicity function. This function could be successfully estimated by also estimating the lateral track irregularities and giving the Kalman Filter self-updating information about the shape of the conicity function. The low adhesion detection was carried out on a complex nonlinear half vehicle model that included saturating contact force equations. The contact forces could be estimated by considering the half vehicle floating on a set of contact forces. Low adhesion conditions can be implied by the relative magnitudes of these contact forces.