Objectives: To assess the reliability of quality indicators of end-of-life cancer care (QI-EOL) and evaluate the quality of end-of-life cancer care in general wards by using QI-EOL. Methods: A retrospective chart review study was conducted on patients with cancer who died in general wards (n = 104) and the palliative care unit (PCU; n = 200) between September 2004 and February 2006 at a regional cancer center in Japan. Herein, we measured QI-EOL, which was developed to evaluate the quality of end-of-life cancer care based on medical charts. We preliminarily assessed the interrater reliability of QI-EOL and subsequently compared the percentages of QI-EOL documented between settings. Results: The reliability of QI-EOL was assured in 27 of 29 indicators (κ > 0.40 or agreement > 90%). For the reliable indicators, we found wide variation in the percentages of QI-EOL documented, ranging from 0% to 98% in general wards. Thirteen of 27 indicators were significantly less documented in general wards than in PCU. Presence of delirium or agitation was less documented (15% in general wards, 55% in PCU, p < 0.001), although presence of pain (92%, 93%, p = 1.000) and dyspnea (78%, 78%, p = 1.000) were similarly documented. Observation and oral care (22%, 62%, p < 0.001) differed significantly. Patient's (29%, 45%, p = 0.009) and family's (30%, 45%, p = 0.014) preferred place of care were infrequently documented. For psychosocial and spiritual concerns, no significant differences were found. Conclusion: QI-EOL is generally a reliable quality measure instrument. We found the need for improvements of end-of-life cancer care in general wards using the QI-EOL.