Background Computed tomography (CT) can be used to assess the rotational alignment of prosthesis components to identify possible underlying causes of symptomatic total knee arthroplasty (TKA). The use of cone beam computed tomography (CBCT) for the imaging of extremities is relatively new, although it has been widely used in dental imaging. Purpose To assess the intra- and inter-observer reproducibility of CBCT, as well as to validate CBCT for TKA component and periprosthetic bone diagnostics. Material and Methods CBCT scans were performed on 18 patients the day before a scheduled revision TKA, from which the intra- and inter-observer reproducibility were assessed. Component rotation and loosening were evaluated. Perioperative bone defects were classified. Results The inter-observer intraclass coefficient correlation (ICC) for femoral component rotation was 0.41 (95% confidence interval [CI] = 0.12–0.69). For the tibial component, the ICC was 0.87 (95% CI = 0.74–0.94). Intra-observer reproducibilities were 0.70 (95% CI = 0.35–0.87) and 0.92 (95% CI = 0.80–0.97), respectively. The sensitivity for tibial component loosening was 97% and the specificity was 85%. The reliability of bone defect classification was only weak to moderate. Conclusion Two-dimensional (2D) CBCT scanning provides reliable and reproducible data for determining the rotation of femoral and tibial components, while showing minor overestimation of tibial component loosening. CBCT is a promising new tool for the evaluation of symptomatic knee arthroplasty patients, with a substantially lower radiation dose compared to conventional 2D multi-slice CT.