本文探討喪葬禮俗中死亡時應「落葉歸根」、「壽終就寢」的變遷狀況與影響因素。檢視歷年死因資料檔中「死亡地點」、「死亡場所」變項，發現死亡時最好在本地自宅過逝的行為仍然存在於台灣社會，但盛行程度已不如以往。再以multinorminal logistic regression分析影響此風俗的決定因素，發現性別、年齡、婚姻狀況、死亡類型、戶籍等變項，對於死亡時是否死於本鄉鎮自宅具有預測能力，但是此模型在過去預測力較強，越接近現代解釋力就越弱。

This thesis explores changes in the practices of “returning home to die” in funeral and burial rituals and the factors concerned and examines death locations in the causes of death data over the years and discovers that in Taiwan’s society, although there are still people who choose to die at home, more and more people die away from home. It then analyzes the key factors influencing this ritual with multinorminal logistic regression and finds out that some factors, such as sex, age, states of marriage, causes of death and household registration, can be used to predict whether or not a person will choose to spend the rest of his days at home. This method, however, is more effective in predicting past behavior than present ones.