The predominance of upper-limb elbow models have been based on earlier lower-limb motion analysis models. We developed and validated a functionally based 2 degree-of-freedom upper-limb model to measure rotations of the forearm using a marker-based approach. Data were collected from humans and a mechanical arm with known axes and ranges of angular motion in 3 planes. This upper-limb model was compared with an anatomically based model following the proposed ISB standardization. Location of the axes of rotation relative to each other was determined in vivo. Data indicated that the functional model was not influenced by cross-talk from adduction-abduction, accurately measuring flexion-extension and pronation-supination. The functional flexion-extension axis in vivo is angled at 6.6 degrees to the anatomical line defined from the humeral medial to lateral epicondyles. The pronation-supination axis intersected the anatomically defined flexion-extension axis at 88.1 degrees. Influence of cross-talk on flexion-extension kinematics in the anatomical model was indicated by strong correlation between flexion-extension and adduction-abduction angles for tasks performed by the subjects. The proposed functional model eliminated cross-talk by sharing a common flexion axis between the humerus and forearm. In doing so, errors due to misalignment of axes are minimized providing greater accuracy in kinematic data.