In order to assess the impact of variations in stroke care on outcomes, and to make geographic comparisons, the three Community Hospital-Based Stroke Programs in North Carolina, Oregon, and New York, aggregated their data on 4,132 hospitalized stroke patients. Complete demographic data or "Major Profile" were obtained on 2,390 (57.8%) of the 4,132 stroke patients. This includes those patients on whom informed patient and physician consents were obtained during the hospitalization. Of the major profile patients, 1,490 (62.3%) were followed for periods up to one year, 502 (21.0%) were lost to followup and 398 (16.6%) died within the one year followup period. Incomplete demographic data or "Minor Profile" were observed on 1,742 (42.1%) of the 4,132 patients. Minor profile includes those who died before comprehensive interviews were completed or those for whom informed consent for an interview could not be obtained. Of the minor profile group, 813 (46.7%) died in hospital, and 929 (53.3%) were alive when discharged from the hospital. This paper, which describes the programs, data collection procedures, and study cases, also highlights specific issues on stroke diagnosis, risk factors associated with stroke, and the influence of interventions on stroke outcomes. We conclude that: 1) the merging of data on hospitalized stroke cases from rural and urban hospitals in geographically distinct regions can be used in the study of stroke diagnosis, the use of diagnostic tests, and the effect of interventions on stroke outcomes; and 2) these data are consistent with the hypothesis that part of the national decline in mortality from stroke is due to a decline in stroke severity.