This paper re-interprets the precautionary principle as a “social epistemic rule”. First, it argues that sometimes policy-makers should act on claims which have not been scientifically established. Second, it argues that, given how scientists ought to solve “inductive risk” problems, such guidance is required not only under actual conditions, but under any plausible conditions. Third, it suggests that procedural fairness may provide policy-makers with reasons to be very reluctant to act on claims which are not scientifically established. The restriction of precautionary reasoning to contexts of significant environmental or public health disaster may respond to this problem.