Rail fasteners are resilient components that connect the rails to sleepers or track slabs, ensuring the integrity of the track structure. Their dynamic characteristics play a crucial role in the noise and vibration performance of the track. The rail fastening system, as a resilient element (RE), can be described by a [Formula: see text] dynamic stiffness matrix based on the interaction between two rigid bodies and the RE. By applying the principles of momentum and angular momentum to the rigid bodies, an algebraic relationship between the dynamic stiffness of the resilient component and the frequency response functions (FRFs) of the assembly is established. Consequently, the dynamic stiffness of the resilient component can be determined based on the experimentally measured FRFs. Two methods are presented for calculating the dynamic stiffness from the FRFs, called the Complete Method and the Partial Method. The two methods are subsequently applied to the structurally complex DT-III type fastener. To do so, a “rail-like” block and a “sleeper-like” block were purposely designed so that they not only can be fastened by the fastener in the same way as in reality, but can also be treated as rigid bodies for frequencies up to 2000[Formula: see text]Hz. FRFs of the rail-like and sleeper-like blocks were obtained through impact tests, and the dynamic stiffness of the fastener in the frequency range 50–2000[Formula: see text]Hz was obtained using both methods. It turns out that, although the two methods have similarly good repeatability, the Partial Method gives more reliable result; in general the dynamic stiffness of the fastener increases with frequency, but exhibits multiple peaks and dips, indicating distinct modal characteristics of the fastener system. The results also show that there is a significant difference between dynamic stiffness seen by the rail and that by the sleeper.