This study validates two-dimensional (2D) ultrasound measurements of muscle geometry of the human medial gastrocnemius (GM) and investigates effects of probe orientation on errors in these measurements. Ultrasound scans of GM muscle belly were made both on human cadavers (n = 4) and on subjects in vivo (n = 5). For half of the cadavers, ultrasound scans obtained according to commonly applied criteria of probe orientation deviated 15° from the true fascicle plane. This resulted in errors of fascicle length and fascicle angle up to 14% and 23%, respectively. Fascicle-like structures were detectable over a wide range of probe tilt and rotation angles, but they did not always represent true fascicles. Errors of measurement were either linear or quadratic functions of tilt angle. Similar results were found in vivo. Therefore, we conclude that similar errors are likely to occur for in vivo measurements. For all cadavers, at the distal end of GM, the true fascicle plane was shown to be perpendicular to the distal aponeurosis. Using transverse images of GM to detect the curvature of the deep aponeurosis at the distal end of the muscle belly is a simple strategy to help identify the fascicle plane. For subsequent longitudinal imaging, probe alignment within this plane will help minimize measurement errors of fascicle length, fascicle angle, and muscle thickness. © 2009 Wiley Periodicals, Inc.