Benefit-cost analysis (BCA) continues to be the principal tool used by American presidents to guide the discretionary decisions of regulatory agencies under their supervision, and increasingly it is viewed by the courts as an important consideration for agencies to take into account in justifying their regulatory decisions. This paper argues that BCA is properly viewed, not simply as a technocratic planning tool, but as a solution to a principal-agent problem. Specifically, it is intended to test whether an agency can demonstrate that it is acting in the public interest. Viewed in this light, some common analytical practices used by regulatory agencies become questionable. A BCA should not, for example, use an assumption that consumers are irrational to support a claim that coercive regulation is making them better off. Consumer sovereignty is axiomatic in BCA, and an agency that uses BCA to justify its actions must accept individuals’ judgments about their own welfare.