Purpose: Design and validation of a magnetic resonance and computed tomography compatible device capable of applying muscle forces to cadaveric knee joints with high precision. Methods: Load was applied to two porcine knee joints at full extension, five and fifteen degrees of flexion. Five repeatability and five reproducibility trials were performed at each flexion angle. Standard deviations (SDs) of joint angle and load were reported. Results: For repeatability, the maximum SDs for joint angle were 1.26° (flexion), 1.54° (ab/adduction) and 0.90° (in/external rotation). The maximum SDs for joint load were 4.60N (anterior/posterior), 7.36N (medial/lateral), and 42.6N (axial). For reproducibility, the maximum SDs for joint angle were 0.84° (flexion), 0.66° (ab/adduction) and 0.92° (in/external rotation). The maximum SDs for joint load were 6.40N (anterior/posterior), 11.7N (medial/lateral), and 39.7N (axial). Conclusions: This level of repeatability/reproducibility is within intra-subject gait variance. Therefore, this device can be an effective tool for in-vitro testing.