Fratricide, or friendly-fire accidents, account for an overwhelming number of casualties during military operations. Gadsden and Outteridge (2006) examined the varying causes of fratricide across the reported incidents. They noted that misidentification accounted for a significant number of fratricide accidents. If methods to reduce misidentification errors are found, fratricide accidents should decrease as a result (Gadsden et al., 2008; Keebler, Sciarini, Jenstch, Fincannon, & Nicholson, 2008). The purpose of this research is to investigate novel training techniques that manipulate the structure of training rewards to reduce misidentification errors. Participants were trained to identify armored vehicles in one of three two-alternative forced-choice training conditions. Specifically, participants received feedback that emphasized response time, response accuracy, or neutral feedback. The feedback was manipulated using game-like points and sound effects. During training, participants receiving accuracy-emphasized feedback exhibited significantly higher training scores than both the speed emphasized, and control groups. As expected, the participants who received speed-emphasized feedback performed significantly faster than the other groups during training. Interestingly, when participants were later tested with a video armored vehicle identification task without feedback, the participants who received the accuracy-emphasized feedback were still significantly more accurate than the other groups. Future research should further manipulate the accuracy-emphasized reward structure to identify optimal ways to deliver feedback during armored vehicle recognition training