Abstract： Objective To explore the training pattern of periodontal probing practice so as to improve the accuracy of probing depth measurements of undergraduate students. Methods Sixty-two undergraduate dental students in their pre-clinical training course were divided into two groups (31 students in each group):the reform group and the control group. The students in reform group learned and practiced periodontal probing by using transparent gingiva models for 30 minutes, and using conventional periodontal probing models for further 90 minutes in two sessions. The total practicing time was 2 hours. The students in control group learned and practiced periodontal probing using conventional periodontal probing models only for 1 hour of practicing time. After the training, new conventional periodontal probing models were used to evaluate the accuracy of probing depth measurements of the students in the two groups. The probing depths of Ramfjord index teeth were recorded by the students and were compared with the results of an experienced teacher's measurements. Results The percentage of absolutely consistent sites in the reform group (63.4%[708/1116]) was significantly higher than that of the control group (60.0% [670/1116]) (P=0.012). In both reform group and control group, the percentages of absolutely consistent sites of the incisors (reform group:69.1%[257/372], control group:65.9%[245/372]) were significantly higher than those of the molars (reform group 58.9%[219/372],control group 54.3%[202/372])(P<0.01), meanwhile, the percentages of absolutely consistent sites of the buccal-lingual sites (reform group:88.7%[330/372], control group:80.4%[299/372]) were significantly higher than those of the interproximal sites (reform group:50.8%[378/744], control group:49.9%[371/744]) (P<0.01). Conclusions The comprehensive reform of periodontal probing training could improve the accuracy of probing depth measurement of the undergraduate students. During the pre-clinical practice, probing training in the molar areas and the interproximal sites should be reinforced.