Most educational investment is based on untested or partially tested assumptions about the cost-effectiveness of a given course of action. Indeed, the only estimates that have been available have been for the average profitability of each type of education, even though these differ greatly from marginal profitability. This article sets out a new approach to estimating the cost-effectiveness of educational investment. The authors canvassed the views of ten world-renowned educational researchers on the likely impact on students' learning achievements of a set of forty measures generally regarded as desirable for improving primary education, and supplemented the responses received with their own calculations of the cost of each, the aim being to establish an index of cost-effectiveness. On this basis, they concluded that the educational projects implemented in the region have failed to include many of the measures identified as the most efficient, and this has limited the quality of the education provided and its potential contribution to economic success, despite the considerable increase in educational investment by governments and international bodies in the 1990s. The article concludes with a number of recommendations aimed at remedying this situation, which take account simultaneously of the impact and the cost of the different educational measures.