Through a record linkage of the 1960 Swedish Population Census and the 1961-79 Cancer Registry it was possible to analyse the occurrence of melanoma and other skin tumours by occupations, classified as either outdoor, office or other indoor work. Office work as compared to other indoor work was associated with risk of melanoma of the covered, but not the uncovered, parts of the body. It is shown that social class is a confounding factor in such analysis, but the elevated risk of melanoma of covered parts of the body among office workers is not entirely due to their higher social class.