限於過去暴露資料缺乏或不完整，及連續性暴露資料測定的準確性、精密度皆會隨時間的不同而有所改變，導致流行病學的研究上暴露與健康效應之確切關係常無法得到一致的結果。工作暴露矩陣模式由於暴露評估之限制與假設較少，因此在因果型流行病學上用來定量個體累積暴露值的機會，有日益增加之情形。工作暴露矩陣模式建立之步驟，隨著研究主題、研究對象特性不同而不同。本文依據文獻上多項研究之綜合，歸納成四個步驟：(1)暴露群的篩選(2)工作職務與暴露因子建立關聯(3)權重大小之決定(4)建立陽性預測值(5)綜合加權給分訂定暴露指標等。而建立一

　　　　　Due to incomplete information about past exposure data and inconsistent accuracy and precision in the monitoring of data during study periods, there have been some problems in exploring the true relationship between exposure and the corresponding health effects in causal epidemiological studies. Compared with the methods of exposure assessment applied in traditional epidemiological research, the use of a job exposure matrix makes a smaller number of assumptions and suffers from fewer limitations. Therefore , the application of a job exposure matrix in determining the exposure level for a homogenous exposure group or individual has become increasingly popular. In general, there are four steps to develop a job exposure matrix: (1) selecting the exposure group, (2) establishing the link between the work task and the hazards, (3) determining the corresponding positive predictive value (PPV), and (4) prioritizing the weighing factors among the variables and scoring the exposure of the homogenous exposure group or individual. The mitigation of bias and evaluation of validity are also important issues throughout the whole process to avoid over- or under- estimation of the true exposure. This article also reviews the applications of research which used a job exposure matrix to explore the relationship between the occupation-exposure and exposure-disease among various epidemiological domains, including in general population or community-based studies, industry- or plant- specific studies, hypotheses generation, and st udy-specific. Some studies have shown a relatively small difference in the exposure estimate between the job exposure matrix and the actual measurement. The use of the job exposure matrix has been showing promising potential in the study of defining exposure-health effect relationships and the methodology of the exposure estimate.