Objectives: to evaluate the effect of using two different voxel sizes on the accuracy of Cone Beam Computed Tomography images for detection of simulated mandibular condylar erosions.Materials and methods: twelve dry human mandibles (24 condyles) were selected and Simulated erosions were drilled with variable bur sizes at different depths on the five condylar surfaces (superior, anterior, posterior, medial and lateral surfaces). The mandibles were scanned using normal and high resolution protocols (voxel sizes: 0.4mm and 0.2mm respectively). Results: Results of the current study revealed that with the CBCT normal resolution protocol (voxel sizes: 0.4mm), the sensitivity of detection of the simulated defects ranged from 87.05% to 94.96% and the accuracy values from 88.9% to 95.86% among the observers. However, higher values were found with the high resolution protocol (voxel sizes: 0.2mm) where the sensitivity values increased to the range of 93.53% to 97.84% and also did the accuracy values with range of 94.44% to 98.15%.Conclusion: In CBCT Despite the higher sensitivity and diagnostic accuracy values of CBCT high resolution protocol both normal and high resolution protocols (voxel sizes: 0.4mm and 0.2mm respectively) were comparable in detection of the simulated mandibular condylar erosions as verified by the very good and significant intra-observer agreements. However, the moderate to good inter-observer agreement pointed out that the differences in agreement might be due to differences in the observer experience more than the difference in the scan protocol.