This Article defends a system with a greater variation in the number of standards of proof than we currently have as both normatively and descriptively valuable. Standards of proof are mechanisms for allocating the risk of factual error between parties. For example, the heightened “beyond a reasonable doubt” standard in criminal cases reflects an aspiration for a legal system erring more in favor of mistaken acquittals than mistaken convictions. Surprisingly, we then assign the same standard to very different cases under the justification that we accept, or should accept, the same error-distribution for those cases. This Article argues that, however ubiquitous, this justification is twice mistaken. First, it is normatively mistaken. There a...