The typical approach in benefit-cost analysis is partial equilibrium. Thus, a policy’s impacts on other markets are ignored. We discuss partial equilibrium evaluation versus general equilibrium ones. It is shown that the rules coincide when markets are perfect and the considered policy is (infinitesimally) small. If changes in some parameters are discrete, the approaches produce different outcomes, in general. In particular, market-based (Marshallian) demand curves no longer reflects the willingness-to-pay for, say, a change in a price. Therefore, income-compensated (Hicksian) tools must be employed. Greater technical detail is expected here that may be more familiar to graduate students in economics.