This paper develops the theoretical underpinnings of conservation ''supply'' curves (CSCs), and in doing so uncovers several problems with current procedures for their construction. The CSC is shown to be derivable from a production isoquant, and not to be a true supply curve. The traditional algorithm for constructing a CSC from discrete measures is shown to be suboptimal, contrary to prior claims. Omitting conservation measures from consideration can lead to systematic, excessive conservation. The CSC concept is extended from constant-service to constant-utility measures, and an improved approximation is suggested for the cost of conserved energy (CCE) of measures that cause rebound. The appendix provides a formula for CCE that is simple yet more general than the one currently in use, bur shows that even with this generalization, CSCs cannot be constructed for a world with fluctuating energy prices.