This thesis presents the investigation into the fatigue propagation and delamination growth of Fibre Metal Laminates under variable amplitude loading. As explained in the first chapter, the motivation of the research is twofold: first, to obtain a clear understanding and detailed characterization of the failure mechanisms in GLARE under variable amplitude loading (selective and flight load spectra). Second is to obtain an accurate prediction model for fatigue crack propagation in GLARE accounting for fibre bridging, delamination and influence of plasticity. The major concept in this thesis is that the stress intensity at the crack tip in the metal layers of a Fibre Metal Laminates (FMLs) is the factor determining the extension of that crack under cyclic loading. This implies that the stress intensity factor can be described with Linear Elastic Fracture Mechanics, including the contribution of the fibre layers and the with the crack growth associated delamination behaviour. The investigation presented in this thesis, covers the theoretical analysis of the crack growth phenomena and experiments to support and validate the developed prediction model. This investigation has been restricted to through-the-thickness cracks with the same crack length in all metal layers. In the second chapter, the various GLARE grades and lay-ups are defined together with a description of the manufacturing process, quality assurance procedures and fatigue crack growth phenomena. The aspects introduced are the fatigue crack growth in the aluminium layers, controlled by the stress intensity factor at the crack tip, and the delamination of the aluminium and prepreg layers, which occurs in the wake of the propagating crack. The crack opening is constrained by the bridging fibre layers, while the stress redistribution to these fibre layers determines the delamination growth. In addition, this chapter discusses the effects of variable amplitude loading in metals, together with the models developed so far to predict fatigue crack growth. The influence of variable amplitude loading on the delamination growth (i.e. a major phenomenon contributing to the slow fatigue crack growth in FMLs) is discussed in chapter 3. An extensive test programme is presented in this chapter utilizing double and multiple block loads, and flight load spectra to validate the hypothesis that delamination in FMLs is independent of retardation effects under variable amplitude loading. Furthermore, the use of scanning electronic microscopy is presented to study the delamination growth striations on the disbonded fracture surfaces. Chapter 4 presents the effect of variable amplitude loading on delamination shapes. These delamination shapes influence the bridging stresses and the crack tip stress intensity factor. The change in the delamination shape due to variable amplitude loading is understood, and the hypothesis on the reason of this change in delamination shapes has been validated. The use of Digital Image Correlation (DIC) is presented to observe delamination shapes and crack tip plastic zones in-situ fatigue testing. Although a change in delamination shape can be observed, its effect on the bridging stress profile is observed to be negligible. Crack tip plasticity and shear-lip formation both were investigated, as presented in chapter 5. Difference in shear-lip morphology was observed in monolithic metals, adhesively bonded metal laminates and FMLs. Monolithic metals revealed typical shear-lip profiles with the early tensile mode followed by a transition mode and finally the transverse shear mode. For metal laminates (without fibres), the mode transition happened quite late during fatigue crack growth, but the transition length was smaller than those of monolithic metals. In FMLs the shear lip profiles were rather difficult to see with the naked eye or optical microscope. However, the straight appearance of the crack when observed from the specimen side, implied a flat cracked fracture surface without shear lips. Plastic zone sizes are compared in the same chapter. In case of monolithic metals and metal laminates, a big difference is observed between the predicted plastic zone sizes (using Irwin’s relation) and the sizes measured with DIC. However, comparing these sizes for FMLs revealed only a small difference. In FMLs, the plastic zone sizes are observed to be independent of the crack length which corresponds to the constant crack tip stress intensity factors. Chapter 6 presents the development and validation of three different types of prediction models. These models include a linear damage accumulation (non-retardation) model, a yield zone model and a crack closure model. The predictions using these models correlated with experimentally observed crack growth behaviour. Chapter 7 outlines the sub-routine added to the prediction model for post-stretched laminates. This routine has been validated with the data from post-stretched GLARE 1 and ARALL from the literature. Chapter 8 summarizes the conclusions of the investigation. It can be concluded that with the proposed prediction model, the mechanism of crack propagation and delamination growth in GLARE is fully described and understood. The prediction model has been validated with experimental crack growth data, and is considered accurate. In its implemented form, the model has the potential to be extended to other material-, geometrical- and test parameters.