The study reported here aimed to establish whether the stakes of examinations taken by students in the final two years of compulsory education in the UK were associated with degree of self‐reported examination anxiety, and whether examination stakes moderated the anxiety–examination grade relationship. Data were collected from 615 students who were due to take examinations conceptualised as high stakes (a terminal examination), mid stakes (a modular examination), or low stakes (a mock examination). Findings suggested that students reported the lowest levels of anxiety and attained the highest grades in the mid stakes examination. Regression analysis suggested that examination stakes do moderate the inverse anxiety–grade relationship, but the effect for high stakes examinations was not in the expected direction. Results are interpreted in the context of limitations to this study’s design. Factors associated with the different timing of the examinations may have influenced results. Due to design limitations, these findings should only be considered provisional and an attempt should be made to replicate the findings using a more robust design. This study highlights the difficulties with designing studies and collecting data in an applied educational context.