Predictively powerful yet opaque models are increasingly used by governments and private actors to make high stakes decisions in the healthcare system, the judicial system, the labor market, and so on. In response to this phenomenon, I argue for a right to explanation as an important, overlooked constraint on decision-making using such algorithms. I ground the right to explanation in requirements of fairness: the fairness of most institutions partly depends on individuals’ ability to engage in what I call informed self-advocacy, and explanations are a common necessary means for individuals to be able to engage in such advocacy. I also argue for two specific rights protections: free expert consultation, and algorithmic impact statements.