In this study, we investigated the specific contribution of creativity, as assessed by the five creativity subtests of the Aurora Battery, to future academic performance, independently of past academic performance. Specifically, in a sample of 1165 7th grade students (UK Year 8, 48.3% female) we first looked at whether the factorial structure of the five subtests was better explained by a domain-general (i.e., the g-factor of creativity) or domain-specific (i.e., verbal, numerical, figural) model of creativity. Then, in a subsample of 315 students (46.3% females) we estimated a structural equation model to explore the mediating role of creativity between students’ performance on different academic tests, specifically the Key Stage 2 test (KS2), taken almost two years before Aurora, and the General Certificate of Secondary Education exams (GCSE), taken almost three and a half years after Aurora. Results showed that Aurora's subtest scores were better represented by a single latent factor (a general creativity factor) than multiple (domain-specific) creativity factors. Furthermore, this creativity factor predicted individuals’ performance on the GCSE four years after taking Aurora, even after controlling for previous academic performance. In addition, creativity mediated the relationship between the KS2 and GCSE. These results suggest that a domain-general form of creativity contributes to future academic performance above and beyond other academics skills.