Since the founding of the first hospice in the United States in 1974, the number of health care organizations providing hospice services has grown rapidly. In 1978, the U.S. General Accounting Office identified 59 operational hospices [1]. A survey undertaken by the National Hospice Organization (NHO) in 1980 found 235 operational programs and many more actively planning to deliver services. By the summer of 1981, the Joint Commission on the Accreditation of Hospitals (JCAH), in studying the feasibility of a voluntary hospice accreditation program, had 650 responses to a national survey [2]. Finally, the 1981 NHO directory identifies 464 operational "provider programs" as well as 33 functioning state-level hospice organizations with an additional 353 programs in various stages of establishing hospice programs of care [3]. The growth of the movement and the public recognition it has received have catalyzed advocacy of Federal support for hospice services. In 1979, the Congress responded by mandating a study to delineate the implications of inclusion of hospice services in the Medicare program. The Health Care Financing Administration (HCFA) then selected 26 hospices (from an applicant pool of 233) to participate in a two-year experimental program. These demonstration sites receive reimbursement for services provided Medicare beneficiaries not otherwise available under current regulations. The special reimbursement provisions went into effect on October 1, 1980. (See Appendix A: Description of the Hospice Reimbursement Program.) In the spring of 1980, the Robert Wood Johnson Foundation and the John A. Hartford Foundation joined with the Health Care Financing Administration (HCFA) to solicit proposals for a national evaluation of hospice care as a basis for future Federal fiscal policy and legislation. Brown University was selected as the evaluation center by competitive process and the grant was awarded on September 30, 1980. The evaluation employs a quasi-experimental design in which the impact of hospice care (with and without reimbursement) on quality of life and costs are compared to non-hospice (conventional) terminal care. Eight hundred patients and families in 24 comparison sites located in three regional areas (Southern New England, Northern Midwest and Southern California) are expected to participate. Primary data collection began on August 1, 1981. Analyses of differential outcome are performed using standard linear multiple regression and logistic multiple regression with separate models for each comparison group. Effects are tested by separately estimating the specific response variable for the prototype (average) hospice patient for each model.