Background Emergency medicine is a complex setting for healthcare delivery which relies on communication, negotiation, teamwork, trust, and shared dialog. The nature of the work comprises dealing with emotionally challenging situations and acting under uncertainty. For healthcare staff this poses the need to be adaptive and open to change. Psychological safety is an important component of productive teamwork and learning in such contexts. Edmondson's model of team psychological safety highlights factors which contribute to the development of psychological safety for staff groups and the mediating role this has for team performance. Aim The aim of the study was explore the link between psychological safety and improvement work. The research question was: Do the aspects covered in the Edmondson model fully describe healthcare workers' perceptions of psychological safety and are all aspects in the model needed to describe these perceptions during testing of new work procedures in an emergency department?" Methods Using a mixed-method approach we investigated a change programme with interviews, a questionnaire and a workshop in an emergency department of a hospital in the Capital Region of Denmark. Thematic analysis of qualitative data and descriptive statistics of questionnaire data were undertaken. Results Data indicate the Edmondson model is useful to help understand and identify important antecedent and outcome factors during a period of testing new work-flow processes. The model could not capture all aspects in this study's data material, and was updated as a result. The main modifications were explicitly integrating the physical aspects of the work setting into the considerations of psychological safety, the inclusion of an additional antecedent factor relating to perceptions of care quality and adopting bi-directional links between the antecedent and consequence elements in the model. Conclusions Although limited in scale, the study supports Edmondson's model of psychological safety as appropriate in describing many of the dynamics experienced by staff engaged in testing new work process. However, additional factors, not included in Edmondson's model and potential adaptations to the model are proposed.