The study comprises 7,410 male employees from the Norwegian primary aluminium industry, and presents overall mortality as well as cancer incidence. During the period 1953-79, 428 new cases of cancer were observed compared to the expected number of 412.2. The only statistically significant difference for the four investigated plants combined was cancer of the lung with 57 observed against 35.9 expected cases. The increased risk was almost solely confined to two subgroups in processing departments; workers with a short duration of employment and workers with a very long duration of employment in the older plants. These results were difficult to interpret because of the problems in selecting the appropriate reference population for estimation of expected number of cases of lung cancer. Furthermore, the individual occupational histories were incomplete and there was no information on smoking habits. In order to determine whether there is an excess risk for workers with long periods of employment in the newer plants, it is necessary to wait for another 10-20 years, when more conclusive data will become available.