A job-exposure matrix has been applied in a case-control study of lung and bladder cancer on the basis of occupational information abstracted from British death certificates. The expected association between lung cancer and jobs entailing exposure to asbestos was clearly demonstrated (relative risk, 1.5; 95% confidence interval, 1.2-1.9). The effects of three other known industrial carcinogens were not apparent, and reasons for this were discussed. Also included in the matrix were five substances whose carcinogenicity in humans has not been established. Formaldehyde, diesel fumes, and cutting oils were all associated with carcinoma of the bronchus, but the absence of a risk in "high-exposure" occupations was against a causal relationship. Bladder cancer was more common in jobs involving high exposure to printing inks (relative risk, 5.0; 95% confidence interval, 1.0-25.8) and cutting oils (relative risk, 1.5; 95% confidence interval, 0.8-2.8). Use of the job-exposure matrix added considerably to the conventional analysis of cancer risk in individual occupational categories.