It is hypothesized using a public choice framework that societies establish social institutions or promote social customs that benefit those that control the political process. Using this framework, it is suggested that the institutional acceptability and therefore the actual practice of child labor will be more prevalent when the other members of a society gain from the use of child labor. Child labor will not be tolerated when the other members of a society are harmed by it. Therefore, in order to explain the cross-country variation in the prevalence of child labor, it is necessary to understand the conditions under which child labor either harms or improves the welfare of the remaining members of a society. It is demonstrated theoretically using a standard neoclassical production model that the non-child-labor factors gain from child labor when the economy is closed. This gain is not the result of exploitation of the children, since the children are assumed to receive their marginal products; nor is the described gain due to the parents appropriating some of the child's income. As an economy becomes more open to international trade, those gains (from child labor) diminish and even turn negative as the size of the economy increases. The model also suggest that child labor will be restricted to the poorer unskilled labor abundant countries as the non-child-labor factors are made worse off by child labor in capital abundant countries. It is shown empirically, using regression analysis, that the cross-country prevalence of child labor falls with increases in a nation's per capita income, its openness to trade, and its economic size. A history of communism is also shown to be significant in reducing the incidence of child labor. All of these variables are found to be highly significant (at the 99 percent level or above); the Box-Cox procedure is used to determine the optimal functional form for the regression. All the data are for 1996 and were obtained from the World Bank's 1998 World Development Indicators. Given that an open economy reduces the benefits of child labor to the other members of a society, an open economy thereby reduces the society's incentive to allow child labor. Therefore, it is argued that trade sanctions, as a remedy for child labor, may be counter-productive; however, sanctions (incentives) could be effective if they sufficiently harm (benefit) the non-child-labor factors. These results also suggest that as nations become more democratic, the process sets in motion economic changes that will undermine the practice of child labor. Beginning with an oligarchy or a capitalist dictatorship, the rulers would choose (in the sense that their economic gains would be the greatest) to have a closed economy with child labor. As a nation becomes more democratic, there will be pressure to open the (labor-abundant) economy since the masses (largely unskilled workers) benefit from increased trade due to Stolper-Samuelson effects. Once the economy becomes sufficiently open, there are no longer any gains to the other factors of production of having child labor, and thus the implicit support for or tolerance of child labor will decline.