<i>Objective:</i> The objective of this study was to estimate the frequency of postoperative lingual nerve (LN) impairment after the surgical removal of mandibular third molars (M3) and to identify potential risk indicators. <i>Subjects and Methods:</i> A prospective cohort study involving 321 subjects who had 443 mandibular M3 surgically extracted at the University Hospital, Amman, Jordan. Predictor variables were categorized as patient-, anatomy- and operation-specific. The outcome variable was the presence or absence of LN impairment. Bivariate and multivariate analyses were performed to identify independent predictors. <i>Results:</i> The prevalence of temporary LN hypoesthesia was 2.5% and all 11 cases resolved within 6 months. Factors that predicted LN injury by univariate analysis were horizontal and mesioangular tooth angulation, bone removal, tooth sectioning, longer operations, lingual flap retraction and bleeding into the socket. In the multivariate logistic regression model, lingual flap retraction (p = 0.002, OR = 26.790) and bleeding into the socket (p = 0.044, OR = 4.918) were the independent predictors. <i>Conclusion:</i> Operative variables were the main predictors of temporary LN damage.