Mathematical Anxiety is the feeling of fear or dislike when dealing with mathematical rich situations. Although math anxiety seems to be innocent it can seriously affect so the learning procedure, as the future carrier directions. The accurate recognition of math anxiety is very important so for diagnostic purposes as for e-learning systems. This work comes to present an automatic system for the detection of math anxiety based on electroencephalographic (EEG) signals, that are supposed to be more subjective, compared to self-report and psychometric questionnaires, since they cannot be intentionally modulated. For this reason we have gathered multichannel EEG recordings from two groups with different levels of math anxiety (Low and High). From these EEG signals we have extracted 466 features and then using a feature selection algorithm we ended to only one feature that was able to recognize math anxiety with 93.75% accuracy using a Naive Bayesian Tree with 10-fold cross validation.