In this chapter, we apply the molecular epidemiological paradigm of biomarkers of exposure, early effect and susceptibility to causal models of leukaemia and lymphoma. The aim is to enhance the development of biomarkers for use in studying the causes of these haematopoeitic cancers in the general population. Two causal models of acute myeloid leukaemia are discussed in detail: chemotherapy-induced and benzene-induced acute myeloid leukaemia. Specific chromosomal changes found in acute myeloid leukaemia may serve as useful biomarkers of early effect in these models, and genetic variants in glutathione S-transferases, NQO1 and DNA-repair enzymes may serve as useful biomarkers of susceptibility. Several causal models of lymphoma exist in which biomarkers could be developed and validated. These include human immunodeficiency virus (HIV) immunosuppression, families with inherited disorders and workers exposed to petroleum products, pesticides or organochlorines. Biomarkers of early effect could include markers of DNA double-strand breaks and aberrant V(D)J recombination, and susceptibility may be related to polymorphisms in genes controlling DNA repair and immunological status. We predict that biomarkers of susceptibility will continue to be studied in the case-control format, perhaps in large pooled studies, but that for biomarkers of early effect, there will be a move away from the study of diseased populations to the study of individuals 'at risk' in the causal models described above.