Most secondary education in England is comprehensive. However, a small minority of grammar schools have retained attainment-based selection. Since students in these schools achieve high grades, some commentators and policy makers believe that England's grammar schools are more effective than its comprehensive schools, and suggest their expansion. This study evaluates whether attending grammar schools is associated with long-term academic benefits at the post-18 stage. It does so by comparing Higher Education (HE) participation rates in grammar schools with those of comprehensive schools in non-selective areas. The evidence reveals that the assumed HE participation benefits of attending academically selective schools are mixed. Acknowledging the limitations of regression models controlling for pre-existing differences between schools, this study reveals the sensitivity of the estimation results, and the challenge of choosing appropriate baseline variables. Thus, this study calls for more thorough understandings of the mechanism of school effectiveness, and stronger research designs for evaluation.