In a search for clues to previously unrecognised industrial carcinogens the occupational and smoking histories of young and middle aged men with different types of cancer have been compared. The study population comprised men aged 18-54 and resident in the counties of Cleveland, Humberside, and Cheshire (including the Wirral). Within this population 2942 patients in whom cancers were first diagnosed during the period 1975-80 were identified retrospectively from hospital and cancer registration records. Lifetime occupational and smoking histories were then sought from these subjects (or if they had died by proxy from their next of kin), using a postal questionnaire. The overall response rate was 52.1%. Analysis of limited occupational data obtained from the hospital notes of 89% of the patients suggests that no serious bias arose from the incomplete response to the questionnaire. The present paper describes the findings for non-respiratory cancers. Some tumours did not occur with sufficient frequency to warrant formal statistical analysis. Nevertheless, examination of the histories of patients with these cancers showed several interesting occupational clusters. In particular, five out of 29 patients with acute myeloid leukaemia had worked in electrical trades. The more common cancers were studied by statistical techniques. A large number of possible occupational associations were examined, and some will probably have achieved conventional levels of statistical significance by chance. The results should therefore be interpreted with caution, taking into account evidence from other studies and the biological plausibility of suggested hazards. Among the more interesting findings were an excess of bladder cancer in lorry drivers (RR=1.6, CI 1.0-2.4) and in men employed in the manufacture of vegetable and animal oils and fats (RR = 4.8, CI 1.8-12.9).