The number of years a child of school-entry age can expect to remain in school is of great interest both as a measure of individual human capital and of the performance of an education system. An approximate indicator of this concept is the sum of age-specific enrolment rates. The relatively low data demands of this indicator that are feasible to meet in practice for a large number of countries have contributed to its popularity and it has recently become part of the influential composite Human Development Index (HDI). However, the quality of this approximation has to date not been systematically investigated. In this study, the error of this approximation with respect to different interpretations of "school-life expectancy" are investigated using micro-simulation, where the true value is known. The results suggest that using the standard approximation for policy conclusions or evaluation requires more careful attention to both interpretation and approximation error than is generally acknowledged. (Contains 10 figures.)