The purpose of this study was to evaluate the reproducibility of the soft tissue (ST) thicknesses at 31 landmarks using the cone-beam computed tomography (CBCT) images obtained from 20 adult subjects. Four observers carried out ST thickness measurements using Skull Measure software, and the inter- and intra-observer error rates were evaluated. Only five of 31 landmarks showed significant differences in recorded ST thickness between the observers. When excluding inexperienced observers, only one landmark showed a significant difference between the observers. Regarding the intra-observer reproducibility, the ST thickness measurements at three landmarks showed low correlation coefficients. The results of this study indicate that CBCT images can be used to measure ST thickness with high reproducibility. However, some landmarks need to be redefined to reliably measure ST thickness on CBCT images.