Attempts at developing oncolytic viruses have been primarily based on rational design. However, this approach has been met with limited success. An alternative approach employs directed evolution as a means of producing highly selective and potent anticancer viruses. In this method, diverse viruses are grown under conditions that maximize diversity and then passaged under conditions meant to mimic those encountered in the human cancer microenvironment. Viruses which evolve to thrive under this selective pressure are isolated and tested to identify those with increased potency (i.e., ability to replicate and spread) and/or an increased therapeutic window (i.e., differentiated replication and spread on tumor versus normal cells), both of which have potential value but the latter of which defines an oncolytic virus. Using ColoAd1, an oncolytic virus derived by this approach as a prototype, we highlight the benefits of directed evolution, discuss methods to “arm” these novel viruses, and introduce techniques for their genetic modulation and control.