The full exploitation of additive manufacturing (AM) capability strongly relies on understanding the process-structure-property relationships. A multi-scale approach is generally employed to accomplish this demanding task due to its ability to account for the intricate hierarchical microstructure of these materials. This strategy requires a synergistic use of a strong physics-based theoretical background and specific experimental investigations. Amongst all the metallic materials that can be additively manufactured, 316L steel is one of the most widely employed, especially when adopting laser-powder bed fusion (L-PBF). The role of microstructure on the monotonic response of this material was extensively characterised, whereas its cyclic elastoplastic response has rarely been addressed. Similarly, existing computational modelling approaches primarily focus on the monotonic response and often rely on assumptions developed for homogeneous initial microstructures, such as in wrought 316L steel, which may not fully capture the process-dependent nature of the AMed material. Therefore, the present study addresses these gaps by proposing an advanced computational model that incorporates the investigated microstructure evolution of L-PBF 316L steel during cyclic elastoplastic deformation. Microstructural investigations were carried out before and after cyclic deformation on specimens loaded at 0.4% strain amplitude for a preset number of cycles. Several techniques, such as neutron diffraction, EBSD, SEM and TEM, were used to gain information on crystallographic texture, sub-grains dislocation structures (e.g. AM-induced cellular structure and persistent slip bands) and dislocation densities. The experimental results highlight the relevant role of sub-grain dislocation structures and dislocation density on the cyclic hardening-softening behaviour of the L-PBF 316L steel. The evolution of these microstructural features then provided the basis for developing a dislocation-based crystal plasticity finite element method (CP-FEM) model capable of predicting the macroscopic cyclic elastoplastic response. The CP-FEM model can serve as a tool with a twofold purpose: supporting the interpretation of the experimental results and providing a mathematical relationship between microstructure and mechanical behaviour that can be exploited to optimise the L-PBF 316L steel performance.