In a large population-based case-control study designed to generate hypotheses regarding possible associations between cancer sites and occupational exposures, an innovative methodology is utilized to infer the subjects' chemical exposures. It features (1) a probing interview to obtain a detailed portrait of each job the subject has held, (2) exposure coding by a team of chemists and hygienists who review each interview, (3) a coding checklist of nearly 300 occupational exposures including the most common ones, and (4) indication by the coders of the level, frequency, and mode of exposure, as well as of their degree of confidence that the exposure occurred. In making coding decisions, the chemists draw upon their experience and upon consultants and bibliographic sources. Resultant data can be combined to produce a variety of semi-quantitative indices of exposure for epidemiologic analyses.