Occupational surveillance systems, which examine mortality patterns by occupation or industry, may contribute to hypothesis generation and priority setting for occupational cancer research and control. In the City of Torino the Local Population and Family Register records and the Census records are matchable through a computerized record linkage procedure. Since the early 1980's we have incorporated demographic events, occurring among residents, into the Census data source, covering birth, death and migration. This paper deals with the occupational mortality, from 1981 to 1985, of 262, 314 men and 146, 125 women, aged 15-64, who held a job at the 1981 Census. Mortality from 31 selected causes of death, including 22 cancer sites, among individuals employed in specific occupations (64 categories) and economic activities (80 categories) was compared with the mortality experienced by the whole cohort of active people. The surveillance design does not take into account potential confounders such as cigarette smoking, alcohol and previous work history. Despite the short duration of follow-up, some occupation-cancer associations, consistently documented in others surveillance studies, have been detected in our study: lung cancer among motor vehicle drivers (SMR 143, 27 obs), metal molders (SMR 178, 8 obs), welders (SMR 241, 7 obs) and wood workers (SMR 218, 12 obs), leukemias and electrical workers (SMR 367, 6 obs).