This study provides an update on measures of educational attainment for a broad cross section of countries. In our previous work (Barro and Lee, 1993), we constructed estimates of educational attainment by sex for persons aged 25 and over. The values applied to 129 countries over five-year intervals from 1960 to 1985. The schooling figures indicated the fractions of the adult population for whom the highest level of attainment fell into seven standard classifications: no formal education, incomplete primary, complete primary, first cycle of secondary, second cycle of secondary, incomplete higher, and complete higher. Information by country about the typical duration of each level of schooling then allowed us to compute the number of years of attainment achieved by the average person in each country at the various levels and in total schooling. The estimation procedure began with census information on school attainment for males and females aged 25 and over. These data came from individual governments, as compiled by UNESCO and other sources. The census values provided benchmark numbers for a subset of the dates (roughly 40 percent) that we were considering. Missing cells were filled in by using school-enrollment ratios by sex at various levels of schooling. The basic idea is that the enrolled population is the flow that adds over time to the prior stock of schooling to determine the subsequent stocks. In this manner, full estimates of educational attainment were obtained at five-year intervals for most countries from the benchmark figures for one or more years and from the reasonably complete data on school-enrollment ratios. The schooling figures that we constructed have clear advantages over the educational variables that have been used in many previous cross-country investigations of the effects of schooling. These studies have typically re-lied on school-enrollment ratios or adult literacy rates, concepts that do not correspond to the stock of human capital that influences current decisions about fertility, health, and so on. For these reasons, the new estimates have al-ready been used in many empirical studies.'Despite these advantages, the data assembled in our previous study had a number of shortcomings. First, the available census in-formation motivated an initial concentration on the adult population aged 25 and over; that is, the data were most plentiful for this age group. For many developing countries, how-ever, a large portion of the labor force is younger than 25. For that reason, the present study provides estimates of school attainment for the population aged 15 and over. Our previous fill-in procedure used the UN's readily available figures on the gross schoolenrollment ratio, the ratio of all persons enrolled in a given level of schooling to the population of the age group that national regulation or custom dictates should be enrolled at that level. For example, the total registered students in primary school are typically compared with the population aged 6-11 years. A tendency for students to repeat grades or to return after previously dropping out means that the gross ratio will overstate the accumulation of human capital. One indication of