Advanced design methodologies enable lighter and more reliable composite structures or components. However, efforts to include fatigue delamination in the simulation of composites have not yet been consolidated. Besides that, there is a lack of a proper categorization of the published methods in terms of their predictive capabilities and the principles they are based on. This paper reviews the available experimental observations, the phenomenological models, and the computational simulation methods for the three phases of delamination (initiation, onset, and propagation). It compiles a synthesis of the current state-of-the-art while identifying the unsolved problems and the areas where research is missing. It is concluded that there is a lack of knowledge, or there are unsolved problems, in all categories in the field, but particularly in the category of computational methods, which in turn prevents its inclusion in the structural design process. Suggested areas where short-term and midterm research should be focused to overcome the current situation are identified.