Background Claims data from Medicare or other payers might not generalize to other populations regarding service use after stroke especially among younger patients. However, high agreement between self-report and Medicare claims data would support the use of self-reported healthcare utilization data in these populations. Methods A population-based sample of 147 stroke participants with traditional fee-for service Medicare and their family caregivers was examined. Concordance with Medicare claims was examined for stroke participant self-report for Emergency Room visits, hospitalizations, and physician visits for a six-month period after stroke, and for both stroke participant and caregiver reports of receipt of Physical Therapy (PT), Speech and Language Pathology (SLP), or Home Health Agency (HHA) visits. Results Agreement was good for Emergency Room visits (kappa 0.75), hospitalization (kappa 0.70), and physician visits (Prevalence Adjusted Bias Adjusted Kappa [PABAK] 0.69) but more moderate for physical therapy, speech and language therapy, and home health agency visits (kappa 0.56-0.63). Caregiver agreement with Medicare claims was similar to stroke participant agreement. African Americans were less likely to self-report therapy compared to whites (OR 0.32 PT, 0.38 SLP, 0.29 HHA, p < 0.03). Younger stroke participants reported lower levels of Emergency Room visits than claims (OR 0.81, p = 0.001). Conclusion Healthcare utilization after stroke can be reliably assessed from Medicare claims, Stroke participant, or Caregiver report for salient events such as hospitalizations and Emergency Room visits. Self-report and caregiver report appear to be less reliable for identifying use of therapy or home health services. Caution should be used when interpreting disparities based on self-report data alone in these areas.