The impact of academically selective schools on children's learning achievement has received global attention for decades. Despite the persistence of early-age selection in many countries, evidence of its impacts is mixed. This study analysed national achievement data covering 149,072 secondary school students and examined whether academically selective schools in England (grammar schools) are superior to their non-selective counterparts in improving students' learning achievement. Unlike previous studies which have either focused on overall national patterns, or emphasised a single local area with limited generalisability, this study evaluates the impacts of grammar schools in each of England's selective local authorities (LAs). Although national evaluation showed that students in grammar schools achieved higher than the counterparts in non-selective schools after controlling for pre-existing differences, the geographical difference is substantial. Meanwhile, although the evaluation does not find a systematic trade-off between the effectiveness of grammar schools and the degree of selectivity within the area, LAs with consistently positive grammar school effects are those with fewer grammar schools and have harder selection requirements. Overall, although the application to grammar schools is usually a highly competitive process, this does not guarantee academic returns at later stages, and the assumed positive effect of early-age academic selection in raising students' learning achievement is doubtful.