死因統計是公共衛生計畫與評估最常使用的參考數據。為了瞭解為什麼臺灣重要的盛行疾病高血壓不是主要死因，本文首先介紹死因統計製表所依據的原死因理念及原死因選擇規則。選擇規則中的連結規則是造成高血壓不易成為原死因的主要理由，該規則規定當高血壓與缺血性心臟病、腦血管疾病、腎炎徵候群同時出現於死亡證明書上時，原死因不選高血壓。本文接著回顧過去死因譯碼過程的信效度研究，研究顯示國與國間譯碼差異的可能原因為：各國死因統計資料收集過程的差異(死因診斷格式差異、開具醫師的資格要求、醫院死亡與非醫院死亡之比率、對屍體解剖的要求 、查證的要求不同)；各國醫師填寫死因診斷的習慣不同、不同國家譯碼員對疾病因果關係的解釋不同、原死因選擇規則本身有模糊不清與自相矛盾的地方。多重死因製表與電腦化原死因選擇系統是目前國際上解決前述過程所衍生問題的方法。

　　　　　Cause-of-death statistics are the most highly used reference in public health planning and evaluation. To understand why hypertension, a highly prevalent disease in Taiwan, is not the leading cause of death, this article first introduces the concept of underlying cause-of-death (UCOD), and the selection rules that guide the tabulation of cause-of-death statistics. The linkage rule of the selection rules is the main reason that hypertension would not be selected as the UCOD. According to this rule, when a d eath certificate lists hypertension with mention of ischemic heart disease, cerebrovascular disease, and nephritic syndrome, hypertension would not be selected as the UCOD. This article also reviews previous studies on the reliability and validity of UCOD coding processes. Reasons explaining the variation in UCOD coding among different countries include 1) differences in the procedure of collecting cause-of-death information (e.g., certifier requirements, format of cause-of-death diagnosis, percentage of d eaths in hospitals, autopsy rates, criteria of query); 2) differences in customs of diagnosing cause of death; 3) differences in interpretation of disease causal relationships; and 4) vague and paradox of selection rules. Two solutions to the above-mentioned problems are tabulation of multiple causes of death, and automation of UCOD selection.