AIMS AND OBJECTIVES The aim of this mixed methods study was to develop science-based content for a systematic death education training system based on the needs of adult cancer patients. The study contained two parts: survey development and Delphi survey. First, a small sample test was conducted to check the reliability and validity of the questionnaire. Next, this questionnaire was applied to investigate adult cancer patients' needs for death education. Then, we invited experts in the fields of nursing management, clinical medicine, clinical nursing and psychological care to carry out two rounds of Delphi consultations to revise the training content. BACKGROUND Death education is often combined with hospice care, which is based on a clear 6-month survival period. However, the survival of adult cancer patients has improved with improved cancer diagnoses and treatments, and death education should be initiated before the dying stage. At the same time, patients' needs for medical information become increasingly important in daily clinical practice. Therefore, a death education programme based on adult cancer patients' needs was developed to help these patients reflect on the meanings of life and death. DESIGN A mixed methods study. METHODS During the survey development period (from April 2017-September 2017), a small sample test (n = 150) was conducted to verify the reliability and validity of the questionnaire on death education needs of adult cancer patients. This questionnaire was developed based on a literature review and discussion among the study group. Next, 324 adults with cancer, recruited from a three-level cancer hospital in Tianjin, China, were surveyed to analyse their needs for death education, using the questionnaire that had been tested in the pretest period. Finally, a Delphi survey was conducted from October 2017-January 2018. A panel of experts (n = 23) recruited from major hospitals, nursing schools and universities in China in the fields of clinical nursing, nursing management, clinical medicine and psychological care took part in the study to revise the training contents based on the investigation results. RESULTS The reliability and validity of results based on the small sample test revealed that the Cronbach's alpha coefficient and the half-degree of reliability of the questionnaire were 0.924 and 0.951, respectively. This demonstrated that the questionnaire had high reliability. The KMO was 0.756 and the Bartlett Test of Sphericity showed p