Objectives: To assess whether dimensional changes occur as shape distortion (unevenly), contraction or magnification (evenly) in cone beam computed tomography (CBCT) considering materials, anatomical regions, and metal artefact reduction algorithms (MAR). Methods: Four cylinders of amalgam (Am), cobalt-chromium (Co-Cr), gutta-percha (Gu), titanium (Ti) and zirconium (Zi) were inserted inside a polymethylmethacrylate phantom in anterior and posterior regions for acquisitions in Picasso Trio and OP300 with MAR enabled and disabled. Two observers measured the dimensions of each cylinder in three axes: Y (height), Z (antero posterior diameter), and X (latero-lateral diameter). Repeated measures ANOVA with Tukey post-hoc test compared the data (α = 5%). Results: Shape distortion occurred for all materials in anterior region of Picasso Trio without MAR (p < 0.05). With MAR enabled, Gu and Ti contracted, while the others showed distortion (p ≥ 0.05). In posterior region, all materials distorted in both MAR conditions (p < 0.05), except Gu, which magnified without MAR (p ≥ 0.05) and contracted unevenly with MAR (p < 0.05). In anterior region of OP300, all materials magnified without MAR, (p ≥ 0.05), and had shape distortion with MAR (p < 0.05). In posterior region, only Am showed magnification without MAR (p ≥ 0.05), while all materials presented shape distortion with MAR (p < 0.05). Conclusion: Dimensional changes of high-density materials in CBCT can be either a magnification, a contraction or a distortion; the last condition is the most prevalent. Furthermore, changes differ considering material, anatomical region and MAR condition.