A case-referent study nested in a cohort was used to evaluate occupational variables in the incidence of breast cancer among nurses. There were 59 cases and 118 randomly chosen referents. The participation rate was 97%. Odds ratios (ORs)and 95% confidence intervals (CIs)were calculated and the weights of potential confounding factors estimated by unconditional logistic regression. The odds ratio for breast cancer in a sister was 2.83 (95% CI 1.03-7.81). Specialization in pediatric, psychiatric, general (surgical and medical), geriatric, and primary care nursing, and "other kinds of nursing," gave an adjusted OR of 1.95 (95% CI 0.84-4.54). When working in different wards was accounted for, the highest adjusted ORs were found among nurses handling cytotoxic drugs, OR 1.65 (95% CI 0.53-5.17), and among pediatric nurses, OR 1.47 (95% CI 0.63-3.41); the lowest ORs were found among nurses in primary health care, OR 0.44 (95% CI 0.20-0.96). Analyses of the data stratified on age showed similar results. Occupational risks were not ascertained. Not only occupation but career-related life-styles should be taken into account in studies of health outcomes among working women.