Many Sub-Saharan African countries have instituted free primary education policies, and this has led to a significant increase in the primary school enrollment rate. However, many children who are in school are not learning. It is not clear whether free primary education policies have contributed to the decline in the quality of education and whether these learning effects are long-lasting. This paper addresses the latter question and estimates the long-term effects of free primary education on educational achievement in Lesotho where the program was phased-in on a grade-by-grade basis, beginning with grade one in 2000. The timing of the implementation created changes in program coverage across age (and grade) groups over time. A semiparametric difference-in-differences strategy is employed that exploits these variations to identify the long-term effects of the free primary education policy on educational achievement, using university examinations records data for student cohorts with and without free primary education. The results indicate that the effect of free primary education on academic performance is bounded between 2 and 19 percentage points, implying that the program increased enrollment without hurting education quality.