Silk embodies outstanding material properties and biologically relevant functions achieved through a delicate hierarchical structure. It can be used to create high-performance, multifunctional, and biocompatible materials through mild processes and careful rational material designs. To achieve this goal, computational modeling has proven to be a powerful platform to unravel the causes of the excellent mechanical properties of silk, to predict the properties of the biomaterials derived thereof, and to assist in devising new manufacturing strategies. Fine-scale modeling has been done mainly through all-atom and coarse-grained molecular dynamics simulations, which offer a bottom-up description of silk. In this work, a selection of relevant contributions of computational modeling is reviewed to understand the properties of natural silk, and to the design of silk-based materials, especially combined with experimental methods. Future research directions are also pointed out, including approaches such as 3D printing and machine learning, that may enable a high throughput design and manufacturing of silk-based biomaterials.