To ensure the feasibility of gas turbines, despite rising commodity prices and emission restrictions, an enhancement of both their efficiency and flexibility is necessary. The consequential higher loading of components at high temperature conditions calls for an increased use of damage tolerant design approaches. To still guarantee a safe operation, a sound understanding and reliable estimations for crack growth under service conditions are indispensable. In this paper, the results from several projects in this field conducted at the TU Darmstadt and involved partners are summarized to identify and describe the various influences on crack growth under creep-fatigue and thermo-mechanical fatigue (TMF) loading. The activation of damage mechanisms under TMF loading and interactions between them is dependent on the temperature cycle and the respective load phasing. Depending on the type of loading (force versus strain control), contrary influences of the phase shift on the TMF crack growth rates are found. This can partially be attributed to the differences in mean stress evolution. Crack initiation and propagation under creep-fatigue and TMF conditions are also often connected with significant scattering of initiation sites and crack growth rates. One reason for this nonuniform behavior is the interaction of geometric discontinuities with the microstructure. To investigate the role of the local grain structure for crack initiation and propagation, in situ observation techniques for crack tip movement and local strain fields were applied. Harsh gradients in the local deformation behavior were identified as origins of secondary crack initiation. To describe crack growth under creep-fatigue and TMF conditions, the linear accumulation model “O.C.F.” was developed. It is based on the contributions of fatigue, creep, and oxidation to crack growth per load cycle. This model is capable of reproducing the effects of time-dependent damage, different load ratios, TMF phase shifts, as well as component geometry. Substantial advantages of this method are its independence from empiric correction factors to assess changing load cycle forms and the possibility to give analytic estimations without the need of extensive data processing. The model is currently validated for three nickel cast alloys, also including single-crystalline and directionally solidified cast variants, different creep-fatigue and TMF loading scenarios, and crack geometries. The model's linear formulation allows assessing the dominant driver of crack growth at each stage of an experiment. These predictions are compared with fractographic investigations and in situ observations of crack paths to identify the mechanisms of crack growth under different TMF load cycle forms.