Abstract: The basic operations of trains, including traction, braking, and curving, are all dependent on the friction, or creep force, between the wheels and rails. In the railway industry, the terminology of wheel-rail adhesion is usually used to represent such friction. The wheel-rail adhesion is governed by a variety of factors, and when the adhesion level fails to meet the requirements for traction or braking, a low adhesion issue occurs and the wheel-rail rolling-sliding turns into full-sliding very rapidly. As a result, vehicles may become out of control, and the surfaces of the wheels and/or rails can sustain damage due to the resulting large creepages. Such occurrences not only jeopardize operational safety and stability, but also lead to increased maintenance costs, presenting a significant challenge for the industry. In terms of high-speed railways, the low adhesion issue becomes more troublesome due to increasing adhesion requirement, stronger wheel-rail dynamic interactions, and various other contributing factors. This may highlight the importance of wheel-rail dynamic adhesion and the corresponding dynamic low adhesion, while it is not yet clearly understood. Therefore, adhesion tests were carried out using a 1:3 scaled twin-rig apparatus at a maximum line speed of 250 km/h, under different conditions of the third medium, speed and axle load. Some dynamic results, such as wheel-rail force, creepage, adhesion coefficient and creep curve were derived by tests. The fluctuation characteristics of the adhesion coefficient and the “second peak” phenomenon were analyzed in detail and were preliminarily explained through the measured vibration acceleration. The results showed that both wheel-rail force and adhesion coefficient exhibited significant dynamic effects, especially under liquid-lubricated conditions. The fluctuation of the adhesion coefficient did not change significantly with speed under wet conditions, but its significance was emphasized by the decrease in the steady adhesion level. For example, the trough of the adhesion coefficient at the first peak was only approximately 0.020 under wet conditions and at 250 km/h, indicating an increased risk of dynamic low adhesion issues. Comparisons of results under different liquid-lubricated conditions revealed that the average adhesion coefficient increased in the following order: antifreeze, oil, soapy water and wet conditions. Furthermore, the second peak phenomenon occurred only under wet conditions, and the fluctuation of the adhesion coefficient under antifreeze conditions was minimal. In contrast, variations in axle loads of 13, 14.5 and 17 t had limited and non-monotonic effects on the dynamic adhesion curves. These results illustrated the complexity of dynamic adhesion behavior. Finally, the quasi-steady adhesion models and creep surfaces of Polach architecture were constructed based on the test data under the axle load of the CR450 high-speed train, providing an important reference for the application of test results and subsequent analysis.