Data from Demographic and Health Surveys conducted in Ethiopia and Nicaragua shed light on two practical questions that arise when turning "Education for All" goals of high literacy and universal primary school completion into more specific targets. First, how should we measure literacy? Second, how many years of schooling should be universalized? Evidence presented here suggests two conclusions. First, objective literacy measures (based on simple tests) should be preferred to more frequently employed subjective measures (based on opinions of household respondents regarding their own literacy and the literacy of other members of their households), especially where school attainment levels are low, if we are to avoid overestimating literacy rates. Second, if a main goal of primary education is to render most students literate by objective measures, then at least 5 years of schooling should be universalized, even though subjective literacy rates reach high levels after only 3 or 4 years of schooling. © 2005 Elsevier Ltd. All rights reserved.