Directed evolution can produce proteins with novel functionalities through coupled rounds of gene diversification and selection. Traditional methods of directed evolution are well established in prokaryotic systems. The recent ‘viral evolution of genetically actuating sequences’ (VEGAS) system takes advantage of Sindbis virus and its error-prone replication to diversify transgenes of interest under selection in mammalian cells. We find that the VEGAS system has major technical issues precluding its use in directed evolution campaigns. These include a rapid loss of system integrity, an inability to propagate viral particles in fresh cells, a predominance of selection circuit ‘cheaters’, and a lower than reported mutation rate. In this form, the VEGAS system is unsuitable for directed evolution in mammalian cells, but engineering of the virus will likely improve its feasibility. Our work highlights key areas for improvement. We are currently developing and testing alternative viral systems for robust mammalian directed evolution.