In this study we adopted a person-centred approach to examine whether students could be identified in distinct clusters on the basis of their test anxiety and academic buoyancy scores, and whether students' academic performance differed accordingly. We performed a cluster analysis on a sample of 469 secondary school students preparing for high-stakes examinations and we identified five empirically-distinct clusters. Three corresponded to a continuum of high test anxiety/low academic buoyancy, mid test anxiety/mid academic buoyancy and low test anxiety/high academic buoyancy. Two clusters corresponded to students with mid-high test anxiety and mid-high academic buoyancy. Academic performance was highest for students in clusters of low test anxiety/ high academic buoyancy or mid test anxiety/ high academic buoyancy. Performance was lowest for students in clusters of high test anxiety/ low academic buoyancy. These findings show how academic buoyancy may lower threat appraisal in some students and show a performance protective role in others.