Structural optimization relies on precisely known material data and accurate yet computationally efficient damage prediction models. In this regard, non-proportional fatigue represents a major source of uncertainty due to a more complex material behavior. While non-proportional loads are found in a large variety of industries, the associated modeling uncertainties lead to increased levelized cost of energy in terms of wind turbines, an unacceptable condition given the urgency of a sustainable global economy.In the wind energy industry tests on the coupon, sub-component and full-scale level are predominantly based on uniaxial loads. In addition, the specimen quality in these tests does not always match the mass-production quality. This is particularly true for the design-driving adhesive joints of rotor blades, where hand-mixed specimens are the state of the art even though dosing machines are applied in rotor blade manufacture. The numerical uncertainty regarding non-proportional fatigue is thus amplified based on a deficit of experiments with representative specimens.This thesis presents a new concept to both accurately and efficiently predict the non-proportional fatigue life by example of a fiber-reinforced rotor blade adhesive. In order to achieve this, the influence of non-proportional loads on the cycles to failure of the adhesive needed to be identified with high certainty. Therefore, manufacturing-induced defects such as pores or stress concentrations on account of the specimen geometry were minimized, resulting in the first virtually defect-free rotor blade adhesive specimens that are suitable for multiaxial experiments. A detailed multiaxial material characterization in static and fatigue loading conditions revealed several misconceptions in comparison to literature such as a rather ductile material behavior, associated modeling differences of (elasto-plastic) shear stresses, a more representative yield criterion (Drucker-Prager) and S-N model (Stüssi-Haibach).Based on the unique experimental data, it was demonstrated that global rainflow-counted equivalent stresses lead to a good fatigue life prediction for proportional loads, while an over-prediction of the fatigue life of up to two orders of magnitude in non-proportional loads is possible. Critical plane algorithms were calibrated using the new experimental data set and found to be substantially more accurate, but impractical due to an extensive computation time and complicated validation. However, a FFT-based re-proportionalization of the stress state in combination with a S-N-based correction factor allows to use global equivalent stresses again in phase shift-induced non-proportional conditions. This way, accurate fatigue life predictions are possible that are several orders of magnitude faster than the critical plane approach. Although demonstrated with a rotor blade adhesive, the new approach can be used with any equivalent stress criterion and thus for any material when a phase shift is the main source of non-proportionality.