The incorrect determination of the flexion-extension axis of the elbow may result in poor clinical outcomes for patients undergoing ligament repairs, joint arthroplasties, and procedures requiring articulated external fixators. The variability in the selection of the flexion-extension axis based on a visual approximation of bony landmarks on the distal humerus was studied within a group of surgeons. Three surgeons of different experience levels independently selected points through which they believed the flexion-extension axis passed on the medial and lateral surfaces of 23 distal humeri. One surgeon repeated the point selection on three separate occasions. These surgeon points were compared with an axis calculated by use of a computer-assisted method that relied on the centers of the trochlear sulcus and capitellum via digitization of these structures. Relative to the computer-generated flexion-extension axis, the error in the surgeons' selections resulted in a mean frontal plane angle of 1.5 degrees +/- 3.0 degrees valgus (range, 6.3 degrees varus to 9.6 degrees valgus) and a coronal plane angle of 1.6 degrees +/- 3.3 degrees external rotation (range, 8.3 degrees internal rotation to 10.2 degrees external rotation). These results suggest that the use of a computer-assisted surgical technique will allow a more accurate determination of the flexion-extension axis of the elbow, which may lead to improved clinical outcomes in patients treated with ligament repairs or reconstructions, elbow arthroplasties, or articulated external fixators.