ABSTRACT Lower third molar extraction is a common procedure in Oral and Maxillofacial Surgeons routine. However, even though this procedure is frequent, sometimes its course is unexpected. In this scenario, some developed classifications help to predict the difficulty level for such extractions. Objective This study aimed to verify the association of Pell and Gregory classifications I and II and surgical difficulty through lower third molar extractions. Methods Voluntary patients from the Dentistry Course of the State University of Rio Grande do Norte, between 16 and 45 years of age, who met the inclusion criteria, had their third molars classified according to Pell and Gregory. Results A total of 42 elements were extracted. There were no statistically significant differences for the correlation of Pell and Gregory classification with the surgical difficulty considering the variables such as surgery time (p=0.419), osteotomy needs (p=0.428) and number of anesthetic tubes (0.939). As for the need for odontosection, only when comparing the teeth classified as AI and IIA, the difference in favor of AI elements was statistically significant (p=0.008). There were no accidents or complications. Conclusions Pell and Gregory classification was not a good predictor of surgical difficulty, and further studies on this subject are recommended.