This work investigates effects of using scaled 3D objects to train for military vehicle recognition, alliance detection, and identification. Past research has demonstrated that training with stereoscopic imagery and objects can lead to stronger learning outcomes when compared to non-stereoscopic training modalities. Therefore, this study sought to investigate the effects of scaled physical object training compared to two current training methodologies, namely military issued cards and military vehicle renderings from a training simulation. An experiment was designed using 1:35 physical scale models, military issued cards containing line drawings, and computer-based renderings. Testing procedures consisted of displaying photographs of the military vehicles the participants were trained on. Results show physical models led to significantly stronger performance compared to cards and images. Limitations and future work are discussed.