The objective of our study was to demonstrate how the intact human hand can be used as a benchmark for electromyogram (EMG)-based myoelectric control of robotic interfaces (e.g., myoelectric prostheses). Using the intact human hand as a gold standard for control algorithms is attractive because able-bodied participants are widely available, have stereotypical movements, and possess highly refined motor control. We compared within-subjects performance of a real-time virtual posture-matching task between a musculoskeletal model-based EMG controller (model trials) and the human hand (goniometer trials). Goniometer trials had lower (i.e., better) normalized path length (2.0±1.6) and task duration (3.3±3.4 sec) than model trials (4.1±4.3 and 12.3±10.7 sec, respectively; p<0.0001). Though, qualitatively, actual (measured by goniometers) and virtual joint angles assumed similar relative postures during model trials, there was a constant offset between them. Additionally, joint angles were more variable during model trials than goniometer trials. The results quantified the extent to which task performance and movement characteristics were not as good with the EMG controller (in this case, the musculoskeletal model-based controller) as with the gold-standard intact human hand. How EMG controllers compare with intact human hand control can drive and inform controller advancements.Clinical Relevance- The gold-standard intact human hand provides an objective way to decide which EMG control algorithms to translate to clinical robotic interfaces.