It has been suggested that simulating physiological loading of the knee during magnetic resonance imaging (MRI) is a promising technique for assessing soft and hard tissues in the knee joint. We have developed a novel MRI-compatible lower limb loading and positioning device to assess knee biomechanics in a physiologically relevant environment using MRI. The objectives of this study were (1) to evaluate inter- and intra-examiner reliability for using our custom-built loading system to maintain a desired load magnitude and direction during each loading trial and over repeated subject visits and (2) to determine the effect of the applied load on motion of the subject's knee over the duration of a loading trial. The pneumatic-controlled loading system was tested on ten subjects at a compression load of 50% of the subject's bodyweight. Two examiners separately positioned and loaded each subject for three loading trials per visit, repeated for three visits. The primary outcome measure was the magnitude of the primary axial load (proximal/distal force) applied to the subject's foot over a loading trial. Secondary outcome measures included average magnitude of medial/lateral and anterior/posterior forces as well as valgus/varus, flexion/extension, and external/internal moments applied to the subject's foot during a loading trial. Location of center of loading at the foot was also recorded. Primary axial load was found to be maintained to within 44-47% of subject bodyweight. Following load-application, the subject's knee exhibited movement throughout the duration of each loading trial. We found that 61.0% of proximal/distal knee displacement occurred within the first 2 minutes following loading. Between minutes 4 and 12, knee positioning was maintained to within 0.92 mm in the medial/lateral direction and 1.24 mm in the proximal/distal direction. We conclude that our loading device can apply controllable and reproducible loading over repeated trials as well as limit subject motion throughout each trial.