[en] Purpose: To investigate the interfractional variability of respiration-induced esophageal tumor motion using fiducial markers and four-dimensional cone-beam computed tomography (4D-CBCT) and assess if a 4D-CT is sufficient for predicting the motion during the treatment. Materials and methods: Twenty-four patients with 63 markers visible in the retrospectively reconstructed 4D-CBCTs were included. For each marker, we calculated the amplitude and trajectory of the respiration-induced motion. Possible time trends of the amplitude over the treatment course and the interfractional variability of amplitudes and trajectory shapes were assessed. Further, the amplitudes measured in the 4D-CT were compared to those in the 4D-CBCTs. Results: The amplitude was largest in the cranial–caudal direction of the distal esophagus (mean: 7.1 mm) and proximal stomach (mean: 7.8 mm). No time trend was observed in the amplitude over the treatment course. The interfractional variability of amplitudes and trajectory shapes was limited (mean: ≤1.4 mm). Moreover, small and insignificant deviation was found between the amplitudes quantified in the 4D-CT and in the 4D-CBCT (mean absolute difference: ≤1.0 mm). Conclusions: The limited interfractional variability of amplitudes and trajectory shapes and small amplitude difference between 4D-CT-based and 4D-CBCT-based measurements imply that a single 4D-CT would be sufficient for predicting the respiration-induced esophageal tumor motion during the treatment course.