Purpose/Significance To collect and process classical formula medical case data and to provide high-quality training data for intelligent diagnosis and theatment models in the field of traditional Chinese medicine (TCM). Method/Process A framework is developed for the collection and processing of classical formula medical case data in TCM. Methods are proposed for data acquisition, cleaning, and similarity detection from multiple sources, including web resources, printed literatures, electronic medical records, and existing databases.Result/Conclusion A total of 12 499 valid classical formula medical case records are obtained, offering a foundation to support the training of downstream TCM intelligent diagnosis treatment models. The proposed data collection and processing method can provide a reference for big data handling in the field of TCM research.