Why is drug development in CNS so damnably difficult? This topic exasperates me, and I'm not the only one. Just noticed this article from 2013. To the author's reason's, I'd add: animal models predict results in humans more poorly in CNS than in, say, anti-infectives. And side effects show more scatter as you move through Phases I, II and III and work with ever-larger populations in trials, so your risk loading is much higher as you move towards approval.