The theory of second best (Lipsey & Lancaster) shows that the presence of irremovable distortions renders the second-best conditions exceedingly complicated; by satisfying some optimality conditions, an improvement is not ensured. However, the complicated second-best rules are neither optimal nor feasible if informational and administrative costs are taken into account. The simple first-best rules are the optimal feasible in an important class of situations (Informational Poverty), implying that analyses based on first-best assumptions are still relevant for practical policy-making. This is so because, with a reasonable concavity assumption, staying with the first-best rules maximises expected benefit. With more (but not perfect) information, third-best policies are appropriate. Some informal illustrative applications of this third-best theory are provided. In particular, average-cost pricing for public utilities may not be far from the third-best optimum and the necessity to raise government revenue through non-lump-sum taxes need not impose any real distortion.