A well established finding is that the cognitive component of test anxiety (worry) is negatively related to examination performance. The present study examined how 3 self-beliefs (academic buoyancy, perceived control, and test competence) moderated the strength of the relationship between worry and examination performance in a sample of 270 final year secondary school students. Participants completed self-reports of academic buoyancy, perceived control, test competence, and cognitive test anxiety, that were matched with examination grades in English, science, and mathematics. Results showed an interaction between worry and perceived control. Students with higher perceived control performed better at low levels of worry. As worry increased, the differential advantage offered by higher perceived control diminished. At high levels of worry control made little difference to examination performance. Interventions designed to reduce worry may not necessarily improve examination performance unless they also target improved control.