Computational models are mathematical models executed by computer and used to simulate the behavior of complicated systems. When such models are used to analyze biological systems, they are given the added descriptor in silico to emphasize their ancillary role to in vitro and in vivo experiments. The computational approach allows us to analyze components of biological systems, understand complex data, predict system behavior, test hypotheses, and develop new hypotheses. Good computational models should be consistent with both observation and biophysical principle. Ideally, it should be possible to take smaller computational models, each a component of a larger system, and then construct a larger model to represent the complete system. One further criterion of a good computational model is falsifiability, since, by Karl Popper's rationalism, a model that cannot be disproved should be considered unscientific. Models that employ many adjustable parameters or constructions can thus be problematic, as they might conform to a large range of potential observables and thus not be falsifiable.

Computational models used to study cancer metastasis are as varied as the many facets of metastasis. Models have been developed to study the ways in which tumor cells can dissociate from the primary tumor to invade into their local microenvironment. Zaman et al., for example, created a three-dimensional model of tumor cell migration. Their model showed how adhesive forces, propulsive forces, and viscous effects from ligands could influence such movements.