This paper analyzes the lasting impacts of a project aimed at teaching children how to read and write at age 6. Using a Difference-in-Differences methodology, the results show that it is not enough to get children literate at age 6 to secure lasting effects; the instruction process is also an important factor: Only pupils exposed to the Phonics method were able to perform better than controls in a Language exam four years later. The results are robust for a large set of time-varying control variables, including the socioeconomic status of the children, the main variable associated with school performance.