OBJECTIVES To develop review criteria from the Agency for Health Care Policy and Research Stroke Rehabilitation Guidelines, to review chart records from three sites of care, and to evaluate the interrater and intrarater reliability for the chart review. DESIGN A descriptive cross-sectional study using a convenience sample. SETTING Charts for abstraction were obtained from three sites of care home health care, nursing facilities, and inpatient rehabilitation centers. PARTICIPANTS Charts were included in the study from the three sites of care if the following conditions were met: (1) the client's first admission to a rehabilitation setting; (2) the client's care was Medicare reimbursed; (3) the client lived in the community prior to the stroke; and (4) the client was receiving skilled rehabilitation services. MEASURES Review criteria, developed directly from the AHCPR Stroke Rehabilitation guidelines, consisted of 11 global quality criteria representative of comprehensive multidisciplinary rehabilitative care. There were approximately 150 variables, comprised of specific criteria to measure each of the 11 global quality criteria plus comprehensive demographic and client-specific information. RESULTS Results of this study suggest that differences exist in documentation of care across the three sites of care. There was difficulty in obtaining adequate numbers of home health charts. Intrarater reliability, using Cohen's Kappa, was .78 and interrater reliability was .64. CONCLUSIONS Based on chart documentation, there is variability in the process of stroke rehabilitation care across nursing facilities, inpatient rehabilitation facilities, and home health. This variability can be reliably assessed by chart review. This study provides the impetus for future research specifically evaluating the associations between documentation of the processes of care and patient outcomes.