Objective: To study the association of postacute care (PAC) settings and mortality outcome of patients who sustained an ischemic stroke. Design: A retrospective cohort study. Setting: An integrated health care system in northern California. Participants: Patients who sustained an acute ischemic stroke between 1996 and 2004, survived the initial acute care hospital stay, and received PAC services within 14 days of discharge (n -= 16,538) and 61 days of discharge (n = 16,468). Interventions: PAC rehabilitation ranked by resource level, that is, inpatient rehabilitation hospital (IRH), skilled nursing facility (SNF), home health (HH), and outpatient (OP). rehabilitation. Main Outcome Measurements: One-year mortality after acute care hospital discharge. Results: The highest level of PAC services received within 14 days of acute care discharge was IRH for 5.6% of patients, SNF for 48.3% of patients, HH for 18.9% of patients, and OP for 27.3% of patients. The highest level of PAC services received within 61 days of acute care discharge was IRH for 10.9% of patients, SNF for 40.4% of patients, HH for 19.1% of patients, and OP for 29.6% of patients. Cox proportional hazard models showed that patients whose highest level of PAC service was provided by an IRH, through HH, or OP had a significantly better 1-year survival than did those admitted to an SNF. The following factors were associated with a higher risk of 1-year mortality: older age, male gender, African American ethnicity, history of previous stroke, higher Deyo-Charlson comorbidity scores, a longer acute care hospital stay, and hospitalization in one remotely located health service area. Conclusions: In the year after a stroke occurred, the rate of patient survival varied based on PAC rehabilitation services. Age, gender, race or ethnicity, history of a previous stroke, comorbid conditions, and service area also were significantly associated with 1-year mortality after acute care discharge. Further investigation of the differences in mortality among PAC settings is indicated. PM R 2011;3:686-694