Mealtimes are important social events in retirement (RH) and long term care homes (LTC). This manuscript describes the development, refining and scaling of the MSILTC as well as inter-observer reliability. Two facilities provided access to their RH (n~100) and LTC (n~30-45) dining rooms. This observation-based tool captures both frequency and nature of interactions. Mealtime observations were carried out by trained researchers for development (n=13 tables), refinement (n=12 tables) scaling (n=17 tables) and reliability (n= 30 tables). Tablemate and staff level sub scores are calculated considering number of residents at the table and duration of the meal. Statistical analysis using Cohen’s kappa demonstrated that the tool possesses adequate reliability for capturing frequency of interaction among residents and staff [kappa 0.712 and 0.790 respectively]; reliability for nature of interaction was lower [kappa 0.590 and 0.441 respectively]. Construct validity testing is planned to complete the development of the MSILTC.