The appropriate order of measure presentation when testing relationships between math anxiety and performance has not been tested empirically. In the current study, 480 university students were randomly presented or not presented with math anxiety questions at multiple testing stages (pretest [state & trait], midtest [state], and posttest [state]). We tested the impact of anxiety measurement timepoint and found that mean anxiety ratings were significantly lower directly after a math fluency test than at all other timepoints. Across timepoints, relationships between math anxiety and performance were consistent. Participants who completed pretest trait assessments reported higher pretest state anxiety and less effort, but effects were non-significant after Benjamini-Hochberg correction, and their performance was not impacted. There were no other significant effects of prior anxiety measurement. Results suggests that, in most circumstances, researchers can measure state anxiety at any assessment stage without concern for its influence on anxiety ratings or math performance.