In 2015, the U.S. Army identified intelligent tutoring as a crucial resource for effective training of soldiers. Specifically, team training is essential as military missions are usually team-based and require extensive coordination. Intelligent Tutoring Systems (ITS) review actions taken by the user and provide dynamic instructions to teach subject matter to an individual. A team ITS assesses the performance of the teams’ individuals, their overall performance as a team, and the interactions of that team to provide dynamic instructions. While extensive work has been conducted regarding single person ITSs, work regarding team-based ITS is limited. A team ITS is difficult to design as the tutor must account for the actions of multiple individuals and their team interactions. The tutor must teach task skills for completing the objective, and team skills for how a team works to meet the objective. This paper describes the implementation, development and evaluation of a Team Intelligent Tutoring System for military teams. We faced challenges such as defining the appropriate levels of cognitive load and team communication required to be successful. The goal of the work was to evaluate an ITS’s effectiveness in a simple team training scenario, a two-person surveillance task in which participants signaled each other using keystrokes. The scenario was constructed using Virtual Battle Space 2.0 (VBS2), and the tutor was built using the Generalized Framework for Tutoring (GIFT). Sixteen two-person teams were run through the study in one of three feedback conditions (individual feedback, team feedback, or no feedback). Their individual and team performance within the task were assessed. We found that participants in the feedback conditions had fewer extraneous keystrokes in the task than those without feedback.