Previous research on music and mathematics anxiety has relied primarily on self-reports without biological measurements. To address whether these parameters were correlated, we included blood pressure physiological measures, the State-Trait Anxiety Inventory (STAI) and the Mathematics Anxiety Rating Scale (MARS) in our study. One hundred and five psychology undergraduates were assigned to sedative, stimulative and "no music" conditions while completing Cambridge GCE O Level mathematical questions. Anxiety was measured pre-, during and posttest. Results showed that MARS was positively correlated with STAI, but not with the physiological measures. A 3 x 3 mixed ANOVA showed differences between the sedative and no music condition for the measures of STAI and MARS, but not for the physiological measures. Further analyses using t-tests found sedative music to elicit a pronounced decrease in systolic blood pressure and the stimulative music to have minimal effect. To explain these findings and the discrepancy with previous studies, we propose a Perception-to-Physiology model for the effect of music in anxiety.