In the task of defining earthquake design accelerations, different approaches have been used worldwide and recent advances have led to changes in the rationale behind the choice of these coefficients. Considering that the two main objectives of earthquake engineering are to guarantee a certain level of earthquake safety while allowing designs to be feasible from an economic perspective, given the scarcity of resources, this paper reviews and compares different approaches that have been used to define the earthquake design coefficient, starting from the classic one, based on the selection of a fixed return period, and covering other more complex proposals that include, in one way or another, the physical vulnerability of the buildings. For each case, we assess how these two earthquake engineering objectives are addressed and review, from the coding perspective, their advantages and limitations. Our findings indicate that only one of these approaches, denoted as optimal design and proposed in 1976 by Rosenblueth, explicitly accounts for both earthquake safety and construction costs. We conclude that instead of aiming to have a uniform criterion to define the design accelerations (e.g., uniform return period or uniform collapse probabilities), an optimization criterion must be used, despite the challenges it presents from the perspective of building codes.