Building energy simulations have found widespread use as decision-making tools for determining design and retrofitting actions. Despite their popularity, there exists a well-reported issue regarding the numerical treatment of structural thermal-storage components in these models. The optimal means of discretising multi-layer structures is complicated by the different thermo-physical properties, material configurations and boundary conditions encountered within building energy models. This paper addresses this information gap by proposing a methodology that can be universally applied to all multi-layer structures, ensuring accurate predictions while avoiding excessive computational cost. Governing dimensionless quantities of Biot and Fourier numbers are utilised within the discretisation process, making the methodology equally applicable to all materials. The presented methodology also accounts for the configuration of materials within multi-layer structures when assigning discretisation levels, leading to nodes being distributed in accordance with expected thermal gradients. The proposed discretisation methodology has been examined for a number of boundary conditions and wall types with excellent prediction accuracy achieved throughout. Additionally, the utility of resistance-only layers has been explored as a means of increasing computational efficiency. This highlighted the importance of considering both layer position and local thermal properties when simulating multi-layer structures. (C) 2018 Elsevier B.V. All rights reserved.