This paper examines a crisis collaboration exercise and tests whether there is a relationship between participants’ past exercise and professional experiences and their perceived levels of collaboration, learning, and utility (CLU). The study reports on data collected from emergency personnel belonging to a Norwegian maritime agency responsible for maritime safety services. Survey data was collected in conjunction with a 2017 maritime oil-spill collaboration exercise in the southern parts of Norway. The personnel held operative positions during the exercise. Forty-two respondents constituted the final data set. Findings indicated that collaboration exercises have an effect, as the participants experienced moderate levels of CLU during the exercise. However, past exercise and professional experience constituted jointly little of the variance in learning (r2 = 0.19) and utility (r2 = 0.02). The results indicated a possible decoupling between exercise behaviour and behaviour in real crisis work, showing a possible dominance of single-loop learning, and a missing constructive alignment between planned learning activities and outcomes. To enhance perceived levels of learning and usefulness, this study recommends a stronger focus on initial simplicity, variation, constructive alignment, and the inclusion of collaboration elements in the design phases of exercises. Comparable research, preferably using the same design and instrument, is recommended.