Physical restraints are still a common problem across healthcare settings: they are triggered by patient-related factors, nurses, and context-related factors. However, the role of some devices (e.g., bed rails), and those applied according to relatives’/patients’ requests have been little investigated to date. A mixed-method study in 2018, according to the Good Reporting of a Mixed Methods Study criteria was performed. In the quantitative phase, patients with one or more physical restraint(s) as detected through observation of a single index day in 37 Italian facilities (27 long-term, 10 hospital units, =4562 patients) were identified. Then, for each patient with one or more restraint(s), the nurse responsible was interviewed to gather purposes and reasons for physical restraints use. A thematic analysis of the narratives was conducted to (a) clarify the decision-making framework that had been used and (b) to assess the differences, if any, between hospital and long-term settings. The categories ‘Restrictive’ and ‘Supportive’ devices aimed at ‘Preventing risks’ and at ‘Promoting support’, respectively, have emerged. Reasons triggering ‘restrictive devices’ involved patients’ risks, the health professionals’ and/or the relatives’ concerns. In contrast, the ‘supportive’ ones were triggered by patients’ problems/needs. ‘Restrictive’ and ‘Supportive’ devices were applied based on the decision of the team or through a process of shared decision-making involving relatives and patients. According to the framework that emerged, long-term care patients are at increased risk of being treated with ‘restrictive devices’ (Odds Ratio 1.87, Confidence Interval 95% 1.44; 2.43; p < 0.001) as compared to those hospitalized. This study contributes to the improvement in knowledge of the definition, classification and measurement of physical devices across settings.