This master thesis deals with the development of a testing device for high temperature tests on an electrodynamic shaker. The electrodynamic shaker is a high-frequency testing system, which is available at the Chair of Mechanical Engineering at the Montanuniversität in Leoben. This system makes it possible to perform fatigue tests up to the very high cycle fatigue regime in an economically acceptable time. To apply a force, the shaker uses resonance effects of a structure and the resulting large amplification of the acceleration amplitudes. This test system shall be extended by a heating chamber, which enables isothermal tests up to a temperature of 350°C. An essential requirement for the device is to use no additional cooling. Parameter studies with a finite-element-program are used to select ideal dimensions and materials for the fixture. The result is a construction consisting of three parts. A titanium flange and a steel flange are connected by an ceramic spacer. Modal analyses are used to determine eigenfrequencies, eigenmodes and participation factors. This makes it possible to optimize the structural dynamics of the device in order to use it as variably as possible with respect to test frequencies. Gain curves and the stresses in the test cross section are determined with steady state analysis under consideration of Rayleigh damping. By testing on an electrodynamic shaker using the manufactured parts, a difference in resonant frequency between simulation and real system of 6,5% is determined. This is only a minor deviation due to influences such as contact conditions and number of elements in the simulation model that cause the system to become stiffer. The damping ratio of the real system is 50% lower than the value used for the simulation. This difference is due to the difficulty to describe damping losses and the conservative damping ratios of the literature. A lower degree of damping is positive because the maximum gain of the system is higher than calculated in the simulation. Strains are measured by strain gauges and stresses in the cross section are calculated. The dependence of the acceleration applied and the stress level are given by a linear relation. For larger loads, such as those used for fatigue tests, an error between measured and calculated values of approximately 3% is determined. Fatigue tests at two load levels show good correspondence with the results of tests on the existing test setup. The simulations as well as the tests on the real system prove the functionality of the testing device.