The invention discloses a similar Chinese herbal medicine search method based on a probability topic model. The method includes the following steps: transforming Chinese herbal medicine information in 'China Great Pharmacopoeia' and 'Chinese Materia Medica' into digital text through an optical character recognition tool; extracting information such as efficacy, property and flavor and channel tropism of Chinese herbal medicines by using a regular expression, and building a Chinese herbal medicine information base; generating corresponding vector spaces according to efficacy, property and flavor and channel tropism attributes of the Chinese herbal medicines respectively, and regulating the vector spaces of the efficacy according to the probability topic model; and at last, calculating similarity of the efficacy, property and flavor and channel tropism attributes between the Chinese herbal medicines according to cosine coefficients, and generating a Chinese herbal medicine similarity database. A user inputs the name of a Chinese herbal medicine, and a system displays the Chinese herbal medicine and similar Chinese herbal medicines visually in the form of a relational graph through searching in the corresponding Chinese herbal medicine pair similarity information database. By means of the similar Chinese herbal medicine search method based on the probability topic model, relevant Chinese herbal medicines can be searched according to attribute similarity, and the method is of great significance for Chinese herbal medicine study and Chinese herbal medicine informatization.