ProRail Traffic Control would like to have examined what the performance of crossovers for rescheduling is in case of partial obstructions. The need for quantitative information about the performance of crossovers in The Netherlands has grown. The research is intended to fill this knowledge gap by investigating the performance of crossovers in The Netherlands in a quantitative manner for two typical corridors. Therefore, the main research question is stated as follows: What is the performance of crossovers with respect to reliability (benefits) and life-cycle costs when rescheduling takes place? The focus in this research is on IVO switches (crossovers in between stations) that are only used for rescheduling. A crossover is a pair of switches that connects two parallel tracks and enables transfer from one track to the other track. The research question has been explored on two typical corridors in the Dutch railway network: corridor Amsterdam: Amsterdam Riekerpolder aansluiting (Asra) – Amsterdam Bijlmer ArenA (Asb)/Diemen Zuid (Dmnz) and corridor Rotterdam: Rotterdam Centraal (Rtd) – Gouda (Gd). Several methods and tools have been used. First, a disruption analysis of the applied partial rescheduling (VSM: VerSperringsMaatregel) in 2011 has been done. The microscopic application TOON has been used to analyze disrupted train operations. Database MUIS contains manually added reports about rescheduling inputted by traffic controllers. With these two tools, similarities and differences between the applied VSM and the real processes and reports from traffic control have been analyzed. Furthermore, insight into the usage of crossovers during rescheduling has been obtained. To get insight into delays and the propagation, data from the Monitoringsystem has been used. The Monitoringsystem generates couplings between train delays. Second, a simulation study of case study Rotterdam has been performed. The microscopic simulation tool OpenTrack has been used to analyze the rescheduling and crossover usage during partial obstruction between Rotterdam Noord Goederen (Rtng) and Nieuwerkerk a/d IJssel (Nwk) (VSM 25.070). Three scenarios have been analyzed: the reference situation: maintaining crossovers at Nieuwerkerk IVO wissels (Nwki) and Moordrecht overloopwissels (Mdo), scenario 1: a crossover at Capelle Schollevaar overloopwissels (Cpso) instead Nwki and scenario 2: both crossovers at Nwki and Mdo are removed. The scenarios have been evaluated through estimations of passenger-, operator- and infra manager costs. Important conclusions that could be drawn from the analyses are that most disruptions that needed partial VSM were caused by defect trains rather than infrastructure related defects. Furthermore, the largest part of the rescheduling went according to the VSM. Although, some disruptions were too short and therefore rescheduling was partly executed according to the VSM. In both case studies crossovers were frequently used during rescheduling. Especially during the first phase–when trains are trapped–crossovers are needed to quickly remove and prevent trains to enter. This will speed up the implementation of the VSM either if partial or complete VSM is used. Moreover, crossovers were also frequently used to couple and abduct ‘defect’ trains to stations/yards and allow implementation of a VSM at once which is the case of delayed anticipated maintenance. With regard to disturbance of crossovers, crossovers do influence the performance of the railway system i.e. according to ProRail’s Asset Management database crossovers lead to a small number of TAOs. Delays and its propagation depend on the severity of disruption. When disruption was shorter (< 3 hours) no other trains on other corridors within the area of Rotterdam and Utrecht were affected. When disruption took longer (> 3 hours) delays were propagated, affecting other trains on other corridors within the area. Furthermore, it can be concluded from the simulation study that scenario 1: crossover at Cpso instead of Nwki resulted in shorter delays of the still operable SPR 4000-series. However, the simulation showed that adding an additional IC 2800-series did not improve the performance of rescheduling. In case of one disruption per year scenario 2: removing the crossovers by removing the tongue and frog has the lowest total costs (€210,557). Removing both crossovers is more expensive (€214,909) as well as maintaining the crossovers at Nwki and Mdo (€240,422) or replacing the one at Nwki by a new one at Cpso (> €240,422). When more than one disruption per year occurs, remaining crossovers at Nwki and Mdo results in lower costs than removing them. Therefore, it can be concluded that with regard to corridor Rotterdam (six disruptions in 2011) maintaining the crossovers at Nwki and Mdo is the most cost-effective. Finally, some recommendations are given with respect to further research. To get reliable data about the performance of crossovers that are applicable to the whole railway network more case studies should be done and different disruptions should be simulated. Moreover, a network-wide analysis of costs and benefits needs to be done to get reliable results for decision making. With respect to ProRail’s new rescheduling philosophy it is questionable whether it is still sufficient in the long term. New technologies in the railway system such as decision support systems that give optimal rescheduling based on real-time delays are promising. Perhaps this could be explored in advance. Also some recommendations are given with respect to the tools that have been used. Reporting the rescheduling process is arbitrarily done in MUIS if compared with the eventually executed rescheduling process obtained from TOON. This makes analyzing disruptions more complex. Therefore, it might be useful to integrate these two systems. With regard to disturbances of crossovers, data of ProRail Asset Management database did not always match with data of TOON and MUIS reports. It could be beneficial for research purposes to couple these databases. A disadvantage of the Monitoringsystem is that not all couplings are being made. A comparable system, TNV-Conflict would be recommended for follow-up studies since it has very accurate data (to the nearest second) about delays and its propagation. Finally, with the help of user-friendly applications such as a convertor that automatically imports the Infra Atlas in OpenTrack, simulation studies become less time-consuming and therefore more accessible.