Unmanned and robotic systems provide great promise and benefit in a variety of military applications. Although each of the varieties of systems have their own capabilities, one common operational challenge is the cognitive demands placed on operators who must make sense out of information presented to them by the system and communicate that information to others while controlling the system. For example, effective operation of the U.S. Army's Small Unmanned Ground Vehicle (SUGV) requires both procedural skill (e.g., use the hand controller buttons, know the menu options) and tactical skill (e.g., know how to use the system to accomplish the tactical mission, such as searching a room). The purpose of this paper is to document the cognitive demands placed on SUGV operators that impact training. The Task Analysis by Problem Solving method of cognitive task analysis was used to identify the categories of knowledge needed to operate and employ the SUGV and to understand how those categories interact. The analysis indicated that SUGV operators must simultaneously operate the vehicle, execute goals and subgoals, apply military knowledge, apply non-military knowledge, and communicate with leaders. The results suggest various recommendations to improve SUGV training. For example, to better estimate the size and shape of a room, the operator must choose to rotate the vehicle or robot head to view the entire space. Such steps are unique to human-robotic interactions and can substantially increase the mental demands placed on the operator. The need for such deliberate, compensating decisions must be understood and addressed through training to avoid compromised information gathering and to minimize cognitive load. By using the SUGV as a model, lessons learned can be generalized to the training of other unmanned and robotic systems in order to minimize cognitive demands and increase tactical effectiveness.