The formulations of fatigue crack growth prediction are still mostly based on phenomenological models. A commonly used formula in the field of high cycle fatigue is the Paris- Erdogan law. For given experimental conditions (such as temperature, stress ratio or environmental conditions) the parameters C and m have to be experimentally determined and considered as material constants. Thus, for a given material, the fatigue crack growth rate (FCGR) depends only on the applied range of the stress intensity factor. In a threshold region a significant shift in the data of the fatigue crack propagation rate can be observed. The shift is induced by different test specimen geometry. To analyses it the authors will present their own laboratory fatigue crack growth rate test data measured on two different specimens with different levels of constraint and for different steels. It is demonstrated that fatigue characteristics (i.e. C, m and Kth) obtained from different specimen geometries are not only properties of the materials but depends on the specimen geometry.