A distributive policy is one which benefits the constituents of one district, but whose costs are bourne collectively.  This paper scrutinizes the commonly held view that such politics will be chosen inefficiently in representative democracies.  The setting is one in which distributive policies are centrally financed local public goods selected by a legislature consisting of elected representatives from each district.  Three different views of distributive policy making from the political science literature are considered. It is shown that only one of these is capable of generating Pareto inefficient local public goods choices, casting doubt on the idea that distributive policy making is an important source of political failure.  However, all three views can generate choices of public goods that leave open the possibility for changes that pass standard cost benefit tests.