In the last decade, simulation technology has become a familiar feature in the Australian rail industry. This has been driven by a strong organisational intent to improve driver learning, and fuelled by the promise that train simulators enrich training, enhance decision support, and engender skills that transfer positively from the synthetic environment to the actual cab. This paper presents research that sets out to investigate how train simulators are currently being integrated into the rail organisation, and how industry end-users are utilising them to deliver training. Focus groups and observations were undertaken at a number of rail organisations that either owned simulators or were undergoing procurement. The data collected were analysed thematically and showed that, whilst practices for integration were varied, simulators were being extremely underutilised, in spite of their keen uptake. Further, effective application appeared hindered by a plethora of usability issues, concerns over technical reliability and simulator fidelity. The findings revealed one or more disconnects residing between the developer's promise and the organisational intent, which destabilised the process for fully integrating a simulator. This paper disentangles this dynamic by exploring the key obstructions impeding the path of effective simulator application.