Context: Care refusal behaviours in people living with dementia pose significant challenges in residential care settings, impacting care delivery quality and staff wellbeing. While Humanitude methodology offers a relationship-centred approach, empirical evidence for its effectiveness remains limited. Objective: To evaluate the immediate impact of a four-day Humanitude training programme on care refusal behaviours among residents with dementia in memory care units. Methods: A pre-post intervention study was conducted across four memory care units in the United States (October–December 2024). Participants included 47 residents with dementia exhibiting care refusal behaviours and 36 healthcare professionals. The intervention consisted of a standardised 28-hour Humanitude training programme. Primary outcome was measured using the Refusal of Care Informant Scale (RoCIS), with secondary outcomes including the Cohen-Mansfield Agitation Inventory and Neuropsychiatric Inventory Questionnaire. Findings: RoCIS scores showed significant improvement from baseline (M = 7.5, SD = 4.0) to post-intervention (M = 2.9, SD = 2.8; p < .001), with a large effect size (Cohen’s d = 1.33). Secondary measures demonstrated substantial reductions in agitation and neuropsychiatric symptoms, particularly in residents with severe dementia. Limitations: The pre-post design without a control group and short follow-up period limit causal inference and assessment of long-term effectiveness. Findings represent only immediate post-training effects; sustained benefits require further investigation. Implications: Humanitude training demonstrates significant immediate effectiveness in reducing care refusal behaviours in residents living with dementia, supporting its integration into standard care practices in memory care settings.