For the vast majority of U.S. Army Soldiers, the first opportunity to engage a realistic moving target with small arms is in combat. Even Infantry Soldiers and special-skill Soldiers (e.g., Snipers) have very limited opportunities to train realistic moving-target engagements. Current capabilities are limited to targets fixed to rail systems or silhouette targets mounted on pickets that can be walked by Soldiers in a firing-range target pit. Without the opportunity to practice engaging realistic moving targets, the Soldier is not able to develop the correct perceptual and motor tuning to adequately engage live moving targets. One solution for the lack of moving target training capabilities is the use of robotic human-type targets (RHTTs). RHTTs can present a realistic three-dimensional human-sized target that can freely move with semi-autonomous control. Furthermore, RHTTs can be programmed to react to events (e.g., flee after another RHTT is hit) and to move in groups in order to provide more complex training scenarios. Even though RHTTs provide a significant increase in training capability, the realism of the RHTTs will ultimately determine the training effectiveness of the capability. In a sense, the training effectiveness question is a matter of human-robot interaction. RHTTs are designed to emulate human beings moving over terrain, moving in a defined area, and reacting to scenario events. If the RHTTs are perceived as freely moving and acting humans, then training can be optimized. Perceptions of realism, shooting performance metrics, and training capabilities inventories were collected from Soldiers training with one type of RHTT to determine the level of target realism. Overall, the RHTT was determined to provide a realistic representation of human targets. However, several factors detracted from realism in certain scenarios. Both the factors that contributed to and the factors that detracted from realism provide insights for developing more effective RHTTs.