A case-control approach has been used to examine mortality from five cancers--oesophagus, pancreas, cutaneous melanoma, kidney, and brain--among young and middle aged men resident in three English counties. The areas studied were chosen because they include major centres of chemical manufacture. By combining data from 20 years it was possible to look at local industries with greater statistical power than is possible using routine national statistics. Each case was matched with up to four controls of similar age who died in the same year from other causes. The occupations and industries recorded on death certificates were coded to standard classifications and risk estimates derived for each job category. Where positive associations were found the records of the cases concerned were examined in greater detail to see whether the risk was limited to specific combinations of occupation and industry. The most interesting findings to emerge were risks of brain cancer associated with the production of meat and fish products (relative risk (RR) = 9.7, 95% confidence interval (CI) 2.6-36.8) and with mineral oil refining (RR = 2.9, CI 1.2-7.0), and a cluster of four deaths from melanoma among refinery workers (RR = 16.0, CI CI 1.8-143.2). A job-exposure matrix was applied to the data but gave no strong indications of further disease associations. Local analyses of occupational mortality such as this can usefully supplement national statistics.