In many developing countries, children complete few years of schooling and learn little during their time in school. There are many estimation problems that confound attempts to understand the impact of education policies on years of schooling and learning while in school. Recent research has focused on implementing randomized trials to get around estimation problems based on retrospective data. While some useful results have been found, many additional studies are still under way. As these results accumulate it is likely that general conclusions can be drawn, but the evidence to date is too limited to draw general policy recommendations.