To describe the proportion of residents receiving rehabilitation in long-term care (LTC) homes, and scores on activities of daily living (ADL) and falls quality indicators (QIs) before and after change from fee-for-service to an episode of care model; and to evaluate the effect of the change on the QIs. Secondary data were collected from all LTC homes in Ontario, Canada, between January 1, 2011 and March 31, 2015. Variables of interest were the proportion of residents per home receiving physical therapy (PT), and the scores on seven ADL and one falls QI. Retrospective, longitudinal study. All data were extracted from the Resident Assessment Instrument Minimum Data Set. Fewer residents received PT after the policy change (84.6 percent, 2011; 56.6 percent, 2015). The policy change was associated with improved performance on several ADL QIs. However, having a large proportion of residents receive no PT or little PT was associated with poorer performance on two of the QIs measuring improvement in ADLs [No PT: -0.029 (-0.043 to -0.014); -0.048 (-0.068 to -0.027). PT <45minutes per week: -0.012 (-0.026 to -0.002); -0.026 (-0.045 to -0.007); p<.01]. While controversial, the policy and subsequent PT service delivery change appears to be associated with improved performance on several ADL QIs, except in homes where a large proportion of residents receive no PT and low time-intensive PT.