By the early 1980s, epidemiologists had identified many important causes of cancer. They had also proposed the 'multi-stage' model of cancer, although none of the hypothesized events in human carcinogenesis had then been identified. The remarkable advances in cell and molecular biology over the past two decades have transformed the scope and methods of cancer epidemiology. There have been a few new discoveries based purely on traditional methods, and many long-suspected minor risks have been estimated more precisely. But modern epidemiological studies often depend on genetic, biochemical or viral assays that had not been developed 20 years ago.