Educational accountability policies have led to a growth in the use of high-stakes examinations for a number of important educational decisions, including the evaluation of teacher effectiveness. As such, educators are under increasing pressure to raise student test performance. In an attempt to prepare students for a high-stakes exam, teachers often resort to using threat-based messages that focus on the negative consequences of test failure rather than messages highlighting students' ability or expectation for high performance. However, the relative influence of teacher messaging (threat-based or facilitating) under different testing conditions is unknown. The present investigation examined the use of fear and efficacy appeals with 487 university students. Anxiety, motivation, and test performance data were collected during a typical, lower stakes testing situation and a higher-stakes, final course examination. A two-way mixed ANOVA and a mediation analysis were used to examine between (i.e. fear and efficacy appeals) and within (i.e. different testing conditions) subject factors. Results suggest that fear appeals significantly harm student test performance relative to efficacy appeals, even when controlling for the impact of intrinsic motivation on test anxiety. Contrary to prediction, student anxiety did not appear to explain the relationship between fear appeals and lowered test performance. The potential implications of findings are discussed, including the importance of instructional context with regards to teacher instructional practices and student success.