Occupational epidemiology has grown rapidly since the late 1970s. Case-referent studies have become popular, but more recent development relates to analyses of cohort data. Length of follow-up and employment status can now be adjusted for in such analyses. Attention should also be given to "time windows" of relevant exposure, not only in cancer studies. In cross-sectional studies of common diseases, the prevalence rate ratio should be used rather than the currently popular but unintelligible odds ratio as obtained by logistic regression. Exposure assessment should involve measures that would best reveal an existing risk and dose-response relationships. New achievements in molecular biology are currently influencing the development in occupational epidemiology. Not only DNA (or protein) adducts as markers of exposure or early effect, but also the possibilities to use data on metabolic polymorphism to identify genetically susceptible individuals attract interest. Activated oncogenes and inactivated tumor suppressor gene are useful for subspecifying various cancer types so as to obtain more sensitive studies.