This paper describes a programme of tyre tests using a purpose built tyre test rig designed specifically to investigate the behaviour of tyres that are used with Road-Rail Vehicles (RRVs). These vehicles are used extensively by the rail industry to support the maintenance of existing rail lines and new construction works. In many cases, using a vehicle that can only operate on the road, or a rail engine that can only operate on tracks will not provide a suitable or effective solution for the task in hand. RRVs are able to operate on both roads and rail, and as such provide a very flexible solution to the rail industry. When operating on the road, RRVs use pneumatic tyres to control their motion and act to all intent and purpose as a normal road vehicle. When operating on rail lines RRVs use a combination of the pneumatic tyres and standard steel rail wheels to control their motion. The rail wheels maintain the directional stability on the track and the pneumatic tyres provide the tractive force to drive and brake the vehicle. Unlike standard road or rail vehicles, there is to date no predictive engineering practice that allows the use of computer simulation to design and optimise the performance of RRVs when they are operating on rails. Computer tools, such as multi-body systems (MBS) analysis are used extensively to design both road and rail vehicles. For road vehicles a tyre model is needed to represent the behaviour in the contact patch between the tyre and the road. For rail vehicles a model is needed to represent the contact force between the train wheel and the track. In both these applications the behaviour is well understood and over the last half century mathematical models have been developed that allow accurate and useful simulation to support the design of new vehicles and trains. In contrast, RRVs have evolved essentially as modifications to standard road vehicles. While the base vehicle may be very well designed to perform on the road, the performance on rail is based on experience and some testing. There is no up-front science involved in designing a RRV to perform effectively during this very important phase of its operation. In order to develop an accurate model to predict RRV performance on rails it is clear that a model of the behaviour of the tyre when in contact with the rail is needed. To date, no such model exists and additionally the testing needed to generate data that could be used to develop a model has never been carried out. The work described here addresses this with the design and build of a unique test rig that can be used to test tyres on rails for a range of operating conditions, and produce a set of initial results that provide a framework for a future tyre/rail model. The paper concludes with a review of the behaviour measured and provides new insights into how well tyres perform on rails and also where the behaviour differs from the well understood behaviour of tyres operating on roads. The important behaviours that would provide the main parameters for a new empirical tyre model are also identified and discussed.