Traditional analytical epidemiology is directed at identifying the association between risk factors and occurrence of disease by using crude exposure data derived from questionnaires or clinical measures, and taking clinical disease as the end point. With the rapid development in molecular biology and laboratory methods, it is now possible to use biomarkers which are capable of identifying molecular events for epidemiologic research. This improved sensitivity enables us to develop a mechanistic understanding of disease causation: a step closer to the unravelling of the "black box" of traditional epidemiology. Biomarkers may be classified as internal indicators of exposure (biomarkers of exposure), indicators of preclinical adverse effect (biomarkers of effect) or indicators of an intrinsic or acquired susceptibility to disease (biomarkers of susceptibility). Biomarkers provide a better definition of exposure and disease status and consequently they could help to reduce misclassification bias in both exposure and disease, reduce the follow-up time in prospective studies, as well as identify possible interactions between risk factors on disease occurrence. However, a biomarker needs to be validated and its distribution in large populations described before it can be used profitably for aetiologic research. Also, the use of biomarkers in epidemiologic research raises other interesting epidemiological and statistical issues like confounding, effect modification and the analysis of repeated measurements. Molecular epidemiology is a multidisciplinary endeavour which comprises molecular biology, epidemiology and biostatistics. Clearly then, to carry out research in this field profitably, the molecular biologist, epidemiologist and biostatistician must acquire not only expertise in their respective fields, but also an integrated understanding of all three fields. The molecular biologist is not merely a laboratory bench worker; the epidemiologist, a field data-collector and the biostatistician, a number cruncher. They must work together to pry open the "black box" to gain a greater insight into how risk factors operate to initiate disease onset and ultimately to make use of this knowledge base to implement preventive measures.