Abstract Background This study aimed to explore (i) the consistency between physician-rated and bereaved family-perceived intensity of death rattle, (ii) the relationship between intensity of death rattle and the bereaved family’s distress and (iii) the bereaved family’s experience and feelings related to suctioning for death rattle. Methods We used matched data for deceased patients from a prospective cohort study of cancer patients admitted to a palliative care unit, and their bereaved families from a nationwide questionnaire survey in Japan. The intensity of death rattle using Back’s score was evaluated prospectively by physicians and retrospectively by bereaved families. Results In total, 1122 bereaved families answered (response rate: 66.7%). Of these, 297 reported the development of death rattle. The maximum intensity of death rattle evaluated by physicians and perceived by bereaved families was poorly correlated (Spearman correlation coefficient 0.188, P = 0.082). The optimal cut-off point of Back’s score for detecting high-level distress was 1/2, with a low accuracy of prediction (area under the curve 0.62). More than 70% of bereaved families indicated suctioning reduced the intensity of death rattle, made patients comfortable and themselves relieved, whereas a similar proportion felt patients were in distress during suctioning. Families who felt suctioning was gently performed and discussed well whether to do suctioning with health care providers felt less needs for improvement. Conclusions Bereaved family-perceived intensity of death rattle did not correlate to physician-evaluated intensity, and the intensity of death rattle itself seemed to poorly correlate to family distress. Gently performed suctioning based on sufficient discussion with families can help reduce family-perceived patient discomfort.