Currently, the night lighting performance of pavilion-style ancient tower buildings is poor and there are few special studies in this area. In order to improve the current situation of poor night lighting effects of pavilion-style ancient tower buildings and optimize their color temperature to show visual suitability, on the basis of only considering the visual presentation of the color temperature effect, subjective evaluation experiments on the visual comfort of the color temperature of the night lighting under different simulation scenarios are carried out, combining with the actual environmental characteristics of the night lighting around the ancient tower, the presentation of the color temperature and the color temperature recommendation of the lighting manual. In the subjective evaluation experiment, the visual comfort evaluation value is used as the label, and C-SVM support vector machine (SVM) classification algorithm is used to construct the subjective evaluation model. By comparing the classification interface and results of different classification models through the scatter diagram and ROC curve, the best color temperature and brightness performance interval of pavilion-style ancient tower buildings and the mapping relationship between the visual comfort under different color temperature and brightness conditions are obtained. Then, the color temperature performance of the night lighting of the ancient towers is evaluated and predicted. Combining the C-SVM classification algorithm with the problems in the lighting field, this study provides a theoretical basis and guidance for designers to better use lighting methods to restore the essential architectural characteristics of pavilion-style ancient towers in people’s minds, so as to improve the night lighting effect of pavilion-style ancient tower buildings.