Abstract: The evaluation method based on questionnaire survey has limited respondents with low quality, therefore, an evaluation method of express service quality was put forward in this paper through sentiment analysis of massive online reviews by selecting online reviews of express companies SF and ST from dianping.com for experiment analysis. First, the SERVQUAL model of service quality evaluation and the related theory of quality evaluation of logistics service was applied combined with text analysis, to establish an index system of quality evaluation of express service through sentiment analysis. Then online reviews such as capturing, phrasing and marking were preprocessed, and the recall ratio and precision ratio under different feature selection algorithm and different classification algorithm were compared. 614 features were extracted and the useful text were identified by choose IG and SVM as the best combination. Furthermore, the polarity analysis and strength analysis of online reviews were conducted based on semantic similarity calculation of HowNet and adverb classification method. Finally, TF-IDF with the weight of evaluation index was applied to evaluate the express service quality. Compared with the rating scores from dianping.com, the experimental results indicate that the proposed method can effectively and better compare differences between SF and ST in each express service quality index. Besides, the general evaluation scores are in alignment with dianping.com.