<p>Due to the complex and variable microstructures of different porous media, it is still a challenging task to balance the accuracy and efficiency of 3D models using 3D reconstruction methods, which significantly affects the precisions and convergences for 3D reconstruction models-based numerical simulations in various computation fields. This review summaries the 3D reconstruction methods for complex microstructures of porous media from the imaging devices, techniques and applications aspects. In this review, six categories of 3D reconstruction algorithms are primarily concluded including stereoscopic tomography reconstruction, stochastic reconstruction, theoretical reconstruction, hybrid reconstruction, 3D point cloud reconstruction and intelligence learning-based reconstruction algorithms, in which the basic principles, advantages and limitations of these 3D reconstruction methods are compared and studied. With the 3D reconstructed models, numerical studies of mechanical property, cracking behaviors, hydraulic and thermal properties of porous media are concluded. The review provides various practical and accessible 3D reconstruction methods for various properties studies of porous media, which are helpful for numerous engineering fields such as deep energy and resource exploitations.</p>