The Cancer Etiology Branch of the National Cancer Institute hosted a workshop, "Validation of a causal relationship: criteria to establish etiology," to determine whether recent technological advances now make it possible to delineate improved or novel criteria for the rapid establishment for cancer causation. The workshop was held in Washington, D.C., December 11-12, 2003, and participants were among the international leaders in the fields of epidemiology, chemistry, biochemistry, microbiology, virology, environmental and chemical carcinogenesis, immunology, pathology, molecular pathology, genetics, oncology, and surgical oncology. There was a general consensus that the rapid identification of human carcinogens and their removal (when possible) or the establishment of specific preventive and therapeutic measures was the most desirable and effective way to have a rapid and positive impact in the fight against cancer. From a clinical perspective, it may be as important to target initiators, cocarcinogens and promoters, if by removing any one of them tumor growth can be prevented. Future studies should focus on interactions among and between different biological, chemical, and physical agents. Analyses of single agents can at times miss their carcinogenic potential when such agents are carcinogenic only in subgroups of individuals because of their genetic background, diet, exposure to other carcinogens, or microbial infection. Epidemiology, molecular pathology (including chemistry, biochemistry, molecular biology, molecular virology, molecular genetics, epigenetics, genomics, proteomics, and other molecular-based approaches), and animal and tissue culture experiments should all be seen as important integrating evidence in the determination of human carcinogenicity. Concerning the respective roles of epidemiology and molecular pathology, it was noted that epidemiology allows the determination of the overall effect of a given carcinogen in the human population (e.g., hepatitis B virus and hepatocellular carcinoma) but cannot prove causality in the individual tumor patient. Molecular pathology cannot determine the overall impact of a carcinogen in the population but can at times prove causality in the individual tumor patient [such as the detection of high-risk human papillomavirus (HPV) in a cervical carcinoma biopsy]. This is possible when molecular techniques have shown that the agent is required for transformation or malignant growth of human cells (such as antisense HPV strategies showing the requirement for the expression of HPV proteins for tumor cell growth) and when there is supportive experimental animal evidence. Ideally, epidemiology and molecular pathology information together with experimental evidence in animals should be available for the most reliable identification of human carcinogens. All sets of data are not always available, and a rapid identification of human carcinogens is in the best public health interest. Swift validation of a causal relationship when followed by a rapid deployment of preventive and therapeutic approaches should lead to a favorable public health impact (such as hepatitis B virus vaccination to prevent hepatocellular carcinoma).