The presented work illustrates the synthesis of a nonlinear model-based observer for the longitudinal dynamics estimation of a wheelset. This approach offers some significant improvements, such as an enhanced adherence to prescribed brake distances and an advanced usability of the condition-based maintenance of brake systems. The investigation of a single wheelset in a test rig environment enables, for example, the usage of an extended set of sensors to reliably validate the observers. The presented observer design process covers three main steps: identification of the characteristic system disturbances, implementation of the nonlinear observer dynamics, and parametrization of the observer via multicase optimization. With regard to the longitudinal dynamics, the variations of the friction conditions in the wheel–rail interface and in the interface between brake pads and brake disc have a crucial influence. Therefore, a parameter estimator and a disturbance observer are implemented, since these methods allow for a specific consideration of these disturbances. The obtained results prove that the observers accurately estimate the system behavior and provide reliable information on the longitudinal dynamics.