To identify aspects of end-of-life care in the U.S. Department of Veterans Affairs (VA) health care system that are not assessed by existing survey instruments and to identify issues that may be unique to veterans, telephone interviews using open-ended questions were conducted with family members of veterans who had received care from a VA facility in the last month of life. Responses were compared to validated end-of-life care assessment instruments in common use. The study took place in four VA medical centers and one family member per patient was invited to participate, selected from medical records using predefined eligibility criteria. These family members were asked to describe positive and negative aspects of the care the veteran received in the last month of life. Interview questions elicited perceptions of care both at VA sites and at non-VA sites. Family reports were coded and compared with items in five existing prospective and retrospective instruments that assess the quality of care that patients receive near the end of life. Interviews were completed with 66 family members and revealed 384 codes describing both positive and negative aspects of care during the last month of life. Almost half of these codes were not represented in any of the five reference instruments (n=174; 45%). These codes, some of which are unique to the veteran population, were grouped into eight categories: information about VA benefits (n=36; 55%), inpatient care (n=36; 55%), access to care (n=33; 50%), transitions in care (n=32; 48%), care that the veteran received at the time of death (n=31; 47%), home care (n=26; 40%), health care facilities (n=12; 18%), and mistakes and complications (n=18; 27%). Although most of the reference instruments assessed some aspect of these categories, they did not fully capture the experiences described by our respondents. These data suggest that many aspects of veterans' end-of-life care that are important to their families are not assessed by existing survey instruments. VA efforts to evaluate end-of-life care for veterans should not only measure common aspects of care (e.g., pain management), but also examine performance in areas that are more specific to the veteran population.