The U.S. Army Training and Doctrine Command's Army Learning Concept 2015 and Army Training Concept 2025 discuss the requirements for adaptive soldier learning models with flexible training delivery methods. Current Game-Based Virtual Environments (GBVEs) have the ability to provide Military Operations in Urban Terrain (MOUT) training based on the Army's requirements, but only for small unit operations. Existing GBVEs lack the capability to support large numbers of users in the same environment at one time or allow the users to engage in critical thinking. Virtual World (VW) technology offers viable solutions to flexibility and scalability challenges found in traditional simulation-based MOUT training such as room clearing tasks, as well as demonstrated the ability to impart valuable training for such tasks. Previous research indicates less experienced Soldiers benefit from VW training (Lackey, Salcedo, Matthews, & Maxwell, 2014). The evidence suggests a need to empirically explore the impact of VW training for operationally relevant tasks on inexperienced populations. This paper presents the results from the second study in a multi-year series of VW Training Effectiveness Evaluations (TEE). The present experiment investigates performance outcomes and user perceptions of 64 novice Soldiers (e.g., ROTC Cadets) using traditional and VW training methods for a room clearing task. Results indicate significant Pearson's product-movement correlation coefficients between the stress-state survey DSSQ and the workload survey NASA- TLX for each training condition and combined training. The survey results offer insight into performance outcomes for the room clearing task. Furthermore, the results reported herein contribute empirically-derived recommendations for the design, development, and implementation of VW training.