Tooth mode has been proven to be adaptable for the diagnosis of periodontal disease. However, in previous studies of tooth mode computation, model settings, including boundary condition and model structure integrity, were quite different. Whether these settings themselves would significantly affect computation results were unknown, and consistent settings that provide support for the generalization of results are desired. This paper aimed to study this by investigating the effect of three commonly used boundary conditions and model structure integrity on tooth mode computation. Three finite-element models of the mandibular tooth-periodontal ligament-bone complex (TPBC) with different levels of structure integrity were constructed. For each model, tooth modes under three boundary conditions were computed, respectively, using modal analysis. Six order modes of each tooth were extracted. The same order mode shapes of different teeth were the same. Compared to the difference of natural frequency (NF) between healthy and periodontal disease, the difference of computed NF of a tooth caused by different boundary condition settings was slight, and the difference of computed NF of a tooth caused by model integrity was slight. The results suggest that the effect of the three boundary conditions and structure integrity of the TPBC model on mode computation of a given tooth is neglectable. Therefore, to compute the mode of a given tooth, these settings can be selected regarding other issues instead of differences caused by these settings themselves. INDEX TERMS Finite element analysis, modal analysis, tooth mode, boundary condition, model structure integrity.