Objectives The goal of sentinel event (SE) analysis is to prevent recurrence. However, the rate of SEs has remained constant over the past years. Research suggests this is in part due to the quality of recommendations. Currently, standards for the selection of recommendations are lacking. Developing a method to grade recommendations could help in both designing and selecting interventions most likely to improve patient safety. The aim of this study was to (1) develop a user-friendly method to grade recommendations and (2) assess its applicability in a large series of Dutch perioperative SE analysis reports. Methods Based on two grading methods, we developed the recommendation improvement matrix (RIM). Applicability was assessed by analysing all Dutch perioperative SE reports over a 12-month period. After which interobserver agreement was studied. Results In the RIM, two elements are crucial: whether the recommendation intervenes before or after an SE and whether it eliminates or controls the hazard. Applicability was evaluated in 115 analysis reports, encompassing 161 recommendations. Recommendation quality varied from the highest, category A, to the lowest, category D, with category A accounting for 44%, category B for 35%, category C for 2% and category D for 19% of recommendations. There was a fair interobserver agreement. Conclusion The RIM can be used to grade recommendations in SE analysis and could possibly help in both designing and selecting interventions. It is relatively simple, user-friendly and has the potential to improve patient safety. The RIM can help formulate effective and sustainable recommendations, a second key objective of the RIM is to foster and facilitate constructive dialogue among those responsible for patient safety.@en