INTRODUCTION: Poor communication about serious injury in older adults can lead to treatment that is inconsistent with patient preferences, create conflict and strain healthcare resources. We developed a communication intervention called Best Case/Worst Case-intensive care unit (ICU) that uses daily scenario planning, that is, a narrative description of plausible futures, to support prognostication and facilitate dialogue among patients, their families and the trauma ICU team. This article describes a protocol for a multisite, randomised, stepped-wedge study to test the effectiveness of the intervention on the quality of communication (QOC) in the ICU. METHODS AND ANALYSIS: We will follow all patients aged 50 and older admitted to the trauma ICU for 3 or more days after a serious injury at eight high-volume level 1 trauma centres. We aim to survey one family or like family member per eligible patient 5-7 days following their loved ones admission and clinicians providing care in the trauma ICU. Using a stepped-wedge design, we will use permuted block randomisation to assign the timing for each site to begin implementation of the intervention and routine use of the Best Case/Worst Case-ICU tool. We will use a linear mixed-effects model to test the effect of the tool on family-reported QOC (using the QOC scale) as compared with usual care. Secondary outcomes include the effect of the tool on reducing clinician moral distress (using the Measure of Moral Distress for Healthcare Professionals scale) and patients length of stay in the ICU. ETHICS AND DISSEMINATION: Institutional review board (IRB) approval was granted at the University of Wisconsin, and all study sites ceded review to the primary IRB. We plan to report results in peer-reviewed publications and national meetings. TRIAL REGISTRATION NUMBER: NCT05780918.