A cohort of 7020 male pottery workers born in 1916-1945 was identified from all employees in dust-exposed trades, subject to medical surveillance by the Benefits Agency Medical Service of the Department of Social Security in Stoke-on-Trent in the United Kingdom. All but 256 (3.6%) were traced, 1016 (15.0%) had died by 30 June 1992, and death certificates were obtained for 940 (92.5%) of the fatalities - 122 from respiratory cancer. After the exclusion of any recorded asbestos exposure, the proportional mortality ratio (PMR) for lung cancer was 1.22 when calculated against national rates (P < 0.02), but the PMR against local rates was 1.04. Logistic regression analyses based on 75 nested case-referent pairs for which the required information was available showed that lung cancer mortality was dominated by smoking and significantly affected by past asbestos exposure. However, for 47 pairs in which both the case and referent had a history of smoking, there was also significant evidence that the risk was related to duration of silica dust exposure in pottery work but not to radiological score.