We developed an arm model by first creating bone models (humerus, ulna, and radius) based on CT images of human arms; muscle spring models (brachialis, biceps brachii, and brachioradialis) based on magnetic resonance (MR) images; and ligament wire models (17 ligaments that connect bones) based on literature and MR images, and then integrating these models. MR images of a human arm with the elbow flexed at arbitrary angles were obtained. Furthermore, these image data and the simulation results were compared to determine the simulation accuracy. The arm model was also used to analyze the behavior of each bone at ligament rupture. Comparison of the simulation results with MR images of a human arm during elbow flexion revealed a deviation of up to 16.7 mm at the midpoint of the ulnar notch of the radius (distal portion). This could be explained by the following: (1) the slice interval (3 mm) used to obtain MR images of a human arm; (2) influence of the articular capsule; and (3) influence of ligaments around the wrist. When medial collateral ligament rupture and ligament rupture around the lateral collateral ligament were simulated with the model, both bones extroverted and introverted, respectively. The deviation of both bones during medial collateral ligament rupture was greater than that during ligament rupture around the lateral collateral ligament, indicating that the medial collateral ligament was a major contributor to elbow stability. The above results demonstrated that the arm model created in this study simulated human arm movement well.