The complexity of the human brain creates significant, almost insurmountable challenges for neurological drug development. Advanced platforms are increasingly enabling researchers to overcome these challenges, by mimicking key features of the brain's composition and functionality. Many of these platforms are called "Brains-on-a-Chip"-a term that was originally used to refer to microfluidics-based systems containing miniature engineered tissues, but that has since expanded to describe a vast range of central nervous system (CNS) modeling approaches. This Perspective seeks to refine the definition of a Brain-on-a-Chip for the next generation of platforms, identifying criteria that determine which systems should qualify. These criteria reflect the extent to which a given platform overcomes the challenges unique to CNS modeling (e.g., recapitulation of the brain's microenvironment; inclusion of critical subunits, such as the blood-brain barrier) and thereby provides meaningful added value over conventional cell culture systems. The paper further outlines practical considerations for the development and implementation of Brain-on-a-Chip platforms and concludes with a vision for where these technologies may be heading.