Recent progress in assessing exposure in occupational epidemiology studies is reviewed. Traditional methods based on surrogate and qualitative measures of exposure are no longer sufficient for searching for new risks, quantifying risks, and learning about their mechanisms. More sophisticated methods and exposure indices are needed that are aimed at estimating the exposure-response relation. Prospective studies, case-referent studies within cohorts, and community-based case-referent studies applying interviews of the subjects or confirming exposures from work-places are designs favoring exposure assessment. Exposure modeling is expected to improve the quality of estimates in industry-based studies. Job-exposure matrices have proved useful, especially in analyses of large studies, provided that they are applied so that misclassification does not significantly bias the results. Misclassification of exposure should be regularly assessed and controlled in epidemiologic studies. Good documentation of the information used and studies on methodological validity and reliability are needed to develop exposure assessment.