The increasing demand for travelling forces railway operators to use their network at maximum capacity. When unforeseen events occur, the time needed to make the proper adjustments before reaching a gridlocked state is too short. The timetable rescheduling models aim to quickly support experienced dispatchers with various optimal solutions to the problem according to predetermined objectives. The master thesis proposes an ex- tension of the meta-heuristic adaptive large neighbourhood search model from Buschor (2020) to solve the railway rescheduling problem that takes into consideration, in this master thesis, the train capacity constraints. In addition, a new ecient passenger assign- ment is proposed to reduce the computation eort at each iteration. The model is tested on a real railway network with the collaboration of a railway consulting company. The results show a substantial impact of the train's capacity on the passenger assignment and the number of successful trips. Moreover, the passenger assignment algorithm provides an ecient time saving on the computation for the ALNS.