Several studies suggest that work in electrical occupations is associated with an increased risk of cancer, mainly leukaemia and brain tumours. These studies may, however, not be representative if there is a publication bias where mainly positive results are reported. To study an unselected population the incidence of cancer was followed up over a 17 year period (1970-87) in a cohort of 2.8 million Danes aged 20-64 years in 1970. Each person was classified by his or her industry and occupation in 1970. Before tabulation of the data on incidence of cancer, each industry-occupation group was coded for potential exposure to magnetic fields above the threshold 0.3 microT. Some 154,000 men were considered intermittently exposed and 18,000 continuously exposed. The numbers for women were 79,000 and 4000 respectively. Intermittent exposure was not associated with an increased risk of leukaemia, brain tumours, or melanoma. Men with continuous exposure, however, had an excess risk of leukaemia (observed (obs) 39, expected (exp) 23.80, obs/exp 1.64, 95% CI 1.20-2.24) with equal contributions from acute and other leukaemias. These men had no excess risk of brain tumours or melanoma. A risk for breast cancer was suggested in exposed men but not in women. The risk for leukaemia in continuously exposed men was mainly in electricians in installation works and iron foundry workers. Besides electromagnetic fields other exposures should be considered as possible aetiological agents.