This study investigated cancer incidence among Icelandic nurses, with a special focus on breast cancer. Cancer incidence in the study group was compared with that of the general female population. Elevated standardized incidence ratios (SIRs) were seen for some cancer sites, including breast cancer. For all cancers, breast cancer, and selected cancer sites, SIR increased with increasing length of time intervals allowed to elapse before the study period began. The nurses with a shorter employment time had a higher incidence of cancer than those with a longer one. Known confounding factors--parity and age at first birth--do not appear to be a convincing explanation of the excess of breast cancer. The possibility cannot be excluded that the nursing occupation constitutes an independent risk factor for breast cancer.