Corrosion of reinforcement affects anchorage capacity. In this study, four levels of analyses were, for the first time, compared with each other and to tests of naturally corroded beams. In the most advanced approach, three-dimensional non-linear finite element (3D NLFE) analyses employing previously developed bond and corrosion models were carried out. These analyses agreed well with the experiments in terms of crack pattern and maximum load capacity. The next approach consisted of 3D NLFE analyses with a pre-defined bond-slip relation between concrete and reinforcement, resulting in reasonable agreement; however, the anchorage capacity was overestimated and the crack pattern deviated from the experiments. At the next level, the bond-slip relation was used together with a measured available anchorage length, and the anchorage capacity was obtained by numerically solving the one-dimensional differential equation; the results were reasonably close to the experiments. In the most simplified approach, a constant bond stress was assumed together with the available anchorage length measured, which underestimated the capacities. In conclusion, the more advanced analyses provide reliable information regarding the structural behaviour, while the two simplified methods are well suited for use in practice.