Aims: In addition to medical facts, medical students participating in simulation-based training are supposed to acquire general knowledge, e.g. heuristics to cope with critical incidents. While active participation is considered a major benefit of this kind of training, a large portion of students’ time is often spent observing peers acting in the simulator. Thus, we instructionally supported learners with a collaboration script (i.e., a set of scaffolds that distribute roles and activities among learners in group learning situations) during observational phases of a simulation-based training. Our script was designed to help learners focus on heuristics and to facilitate more (inter-)active participation. We hypo-thesised that scripted learners would benefit from the instructional support with respect to individual and collaborative learning processes as well as individual learning outcomes. Methods: Thirty-four medical students in their 7^th to 12^th semester participated in this field study with control group design. The independent variable was the collaboration script (with/without). Four voluntary emergency courses with a full-scale simulator were examined. The acquisition of skills related to Crisis Resource Management (CRM) heuristics was one of the learning goals of these courses. The collaboration script induced learners to perform specific activities during and after each observational phase of the training. Further, the script sequenced the order of activities and assigned roles to the learners. Learning processes were measured on an individual level (by means of notes taken by learners during observational phases) and on a collaborative level (by means of learners’ comments). Learning outcomes were measured with pre-and post-self-assessment of CRM skills and a brief video-based CRM skills test at the end of the course. Results: The collaboration script had the expected positive effect on individual and collaborative learning processes, leading to an increased focus on heuristic strategies and increased collaborative activity of scripted learners. There was no evidence that the experimental conditions differed regarding the objective measure of individual learning outcomes. However, self-assessment data revealed that students in the control condition perceived a higher improvement of CRM skills throughout the course. We suggest that our script might have helped learners adjust an illusion of their own competency – such an illusion may have appeared in the control group as a result of processing fluency. Conclusions: Findings suggest that simulation-based training in medical education can be enhanced with additional instructional support in the form of collaboration scripts designed to turn observational course phases into more active and better focused learning experiences.