Abstract：In this study of affect labeling there are two main questions. The first one is whether affect labeling will enhance or reduce the negative emotion of subjects, especially for Chinese subjects using Chinese characters. The second one is whether using words to describe the non-emotional aspects of the negative faces or pictures will reduce the emotion or not. To test these questions, we conducted two experiments. In the first experiment, 30 female university students participated. The pictures of negative faces (fear and anger) were used as materials and physiological responses, including finger temperature, skin conductance, and finger pulse, were recorded using BiopacMP150 under affect-labeling, affect-matching, gender-labeling, gender-matching, and shape-matching conditions. In the second experiment 28 female university students participated. The pictures of negative scenes were used as materials and the same physiological responses as in the first experiment were recorded using BiopacMP150 under affect-labeling, artificial vs. natural labeling, picture-matching, and shape-matching conditions. The results showed that in the first experiment, the finger pulse rates were lower in the affect labeling condition than in the affect-matching condition, but there were no differences in finger pulse rate between the gender-labeling and the picture-matching condition. In the second experiment, the finger temperatures were lower in the affect-matching condition than in the affect-labeling condition, and they were also lower in the artificial vs. natural labeling condition than in the affect-labeling condition. There were no differences in finger temperatures between the artificial or natural labeling condition and picture-matching condition. In conclusion, the present study demonstrated that affect labeling can reduce the activity of the autonomic nervous system, but that gender labeling and artificial or natural labeling cannot reduce the activity of the autonomic nervous system.