This article presents a protocol for assessing team performance in terms of control based on cognitive systems engineering theory. The protocol changes the focus of team performance assessment from good and bad behaviours towards team activity. By using Hollnagel's Contextual Control Model (COCOM), a protocol is developed so that team activity can be described as four control modes. Data is collected through observation and questionnaires and is analysed in time intervals. Each time interval is then given a control mode. Based on how the control mode changes over time the performance of the team can be assessed. The protocol was tested in a pilot study where commanding staff exercises were analysed. The results show that the protocol can be used to identify differences in team performance. The study also gives empirical validity to COCOM in that the loss of control in emergencies corresponds to an opportunistic or a scrambled control mode.