PURPOSE: Measure associations between clinicopathological and immunohistochemical human Mut-L homologue 1 (hMLH1) gene, and human Mut-L homologue 2 (hMSH2) genes, variables in recurrent AMBs. METHODS: This study consisted of a research retrospective, observational case-control study consisting of 22 cases of recurrent AMB and 22 non-recurrent cases. Cases of AMB with more than one year of follow-up were included in the study. Quantitative immunohistochemical analysis was performed considering the cellular location (nuclear) of the proteins studied. The McNemar test was used to compare variables between primary and recurrent AMBs. Recurrence-free survival was analyzed by the Kaplan-Meier method and survival functions were compared according to the variables using the log-rank test. RESULTS: The posterior mandible was the most affected site in the recurrent (n = 18, 81.8%) and non-recurrent groups (n = 16, 72.8%). Recurrence-free survival was 50.0 (34.5-63.6) months. The following factors were significantly associated with AMB recurrence: presence of cortical bone expansion (p = 0.01), absence of bone reconstruction (p = 0.02), conservative treatment (p = 0.02), loss of hMSH2 (p = 0.01) and hMLH1 (p = 0.04) immunoexpression, and strong Ki-67 immunoexpression (p = 0.03). The risk factors for AMB recurrence were anatomical location (OR = 3.31), locularity (OR = 1.07), cortical expansion (OR = 6.17), cortical perforation (OR = 2.10), bone resorption (OR = 1.52), tooth impaction (OR = 1.86), jaw reconstruction (OR = 6.92), and immunoexpression of hMSH2 (OR = 10.0) and hMLH1 (OR = 4.50). CONCLUSION: Radiographic appearance, treatment modality, and immunoexpression of mismatch repair proteins can be used as predictors of AMB recurrence.