This paper addresses the methodology of cost-benefit analysis as applied to policy alternatives regarding legally sanctioned gambling in its various forms. Existing economic studies regarding the social costs of gambling are reviewed and critiqued. Distinctions are made between definitions of social costs that are defined as actions which result in negative changes in aggregate social wealth (the "narrow" definition), and those which also include internal nonmarket costs that are borne by individual gamblers and their immediate families and acquaintances (the "broader" definition). This distinction is important because of its bearing on economic policies that are primarily concerned with economic efficiency versus policies that are more paternalistic, which attempt to protect individuals from self-damage or self-destruction by restricting their ranges of choice. Whether societies choose to prohibit or severely restrict permitted gambling, or allocate substantial resources to mitigate its negative side effects, rests largely on which of these perspectives regarding social costs they find more appropriate. Finally, the issue of social protection through restrictions on the availability of gambling to the entire population, versus a strategy tailored toward identifiable "problem" gamblers, is discussed.