In this talk a Bayesian approach to trials of rare diseases are discussed. In rare disease situations the ability to do “type I error” controlled trials can be impossible—more patients than exist! There is no single idea that will transform rare disease scenarios—many situations are unique: some have historical data available, some have related diseases, and some have multiple experimental drugs available to test—an even harder problem! In this talk I provide example scenarios with innovations that can be used to provide a way forward. All these approaches include adaptive designs, may include borrowing from historical data, borrowing from other diseases, or running platform trials with multiple experimental treatments.