Measuring emotion usually involves use of questionnaires. However, a questionnaire method is unable to measure emotion in real time as questionnaires are executed after the experiments. Besides, the resolution time of the emotion rating is low. Therefore, we proposed a novel method of measuring emotion in real time and to classify the negative emotion evoked by irritation of traffic jam and Positive emotion evoked when traffic jam disappeared by using bio-signals (electrocardiogram and Pulse wave). We reproduced a situation where traffic jam occurs using a Driving simulator (DS), and we measured bio-signals when subjects drive in it. Measured data were classified using Linear discriminant analysis (LDA) in two states of any three following situations: (Positive or no evoked emotion, Negative or no evoked emotion, Positive or Negative emotion). As a result of classification based on bio-signals, two states of three situations got high classification capability. We also indicated possibility for classification of two states.