This paper evaluates the impact on dropout rates of a policy change in Mexico that eliminates grade retention for all first to third-grade students, causing a sharp reduction in repetition rates. I use a 12-year panel of schools to exploit such variation and estimate Difference-in-Difference models showing an average decrease in dropout rates of 30%. However, this effect is concentrated in wealthier schools, suggesting that social promotion alone is not enough to offset the influence of socioeconomic factors on school attainment. Further evidence shows that eliminating the 'threat' of grade repetition did not reduce average students' performance in standardized tests.