We investigate the impacts of the 2002 elimination of primary school fees in Tanzania. We explore how the magnitude of these effects depends on differences in gender and colonial investment in educational infrastructure. We use 2002 and 2012 census waves as well as historical information on the location of schools in the late 1940s, to conduct a diff-in-diffs analysis. Our results show that exposure to an average of 1.7 years of free primary education reduces the proportion of people who never attended school by 6.8 percentage points. The benefits of fee removal have been significantly larger for females as compared to males, and females from districts where the size of investments in education was larger during colonial rule are the greatest beneficiaries.