•The upright knee rotational alignments were measured by upright computed tomography.•The tibial anteroposterior axis was changed under upright, weight-bearing conditions.•The tibia rotated 7.4° more externally under upright, weight-bearing conditions than under supine non-weight-bearing conditions. The traditional anteroposterior (AP) axis (i.e., Akagi’s line) has been widely used as the tibial component AP axis during total knee arthroplasty (TKA). However, this AP axis has been defined based on computed tomography (CT) in a non-weight-bearing supine position. In this study, AP axes of the tibial plateau from upright CT in weight-bearing and non-weight-bearing positions were determined and compared. This study included 43 knees from 23 healthy volunteers. CT images were obtained in weight-bearing and non-weight-bearing standing positions using a 320-detector row upright CT scanner. The line perpendicular to surgical transepicondylar axis projected onto the tibia plateau was determined as the AP axis in upright weight-bearing and non-weight-bearing conditions. Angular differences between these two conditions were measured. The upright weight-bearing AP axis was positioned in a mean of 7.4 ± 4.3° of internal rotation relative to the traditional AP axis. Distance between the traditional and upright weight-bearing AP axis was 2.9 ± 1.6 mm at the edge of the tibial plateau. The upright non-weight-bearing AP axis was positioned in a mean of 3.5 ± 4.1° of internal rotation relative to the traditional AP axis. Mean angular difference between weight-bearing and non-weight-bearing conditions was 3.9 ± 4.1°. The upright weight-bearing AP axis was positioned in 7.4° of internal rotation relative to the traditional AP axis, showing one-seventh of the tibial tuberosity away from the medial border of the tibial tubercle, which represents a practical landmark for the tibial component AP axis during TKA.