Breast cancer patients who are obese have a higher risk of lymph node metastases and a poorer prognosis than those who are slim. It has been claimed that estrogens derived from fat are important for these associations. If estrogens are important, these relationships must be stronger in the hormone receptor-positive than in the hormone receptor-negative groups. Body mass index (BMI) was used as a measure of obesity. The second, third, and fourth quintiles of BMI were treated as one group and termed 'medium'. Patients in the fifth quintile were termed 'obese' and those in the first quintile 'slim'. The number of women with unilateral disease treated with modified radical mastectomy and included in the study was 1211. Of all patients included, obese patients had a 1.53 higher risk of lymph node metastases compared to slim patients (p=0.02). In the PgR-negative group, obesity gave a 3.08 times higher risk of lymph node metastases (p=0.03). The risk of dying of breast cancer tended to be higher in obese than in slim patients when all patients in the study were compared (relative risk=1.38, p=0.06). BMI did not show a statistically significant relationship with prognosis if only hormone receptor status was considered. However, if lymph node status and hormone receptor status were taken together, the association was strong and reversed in the lymph node-positive group with ER-negative tumours. The adjusted relative risk was 0.33, showing that slim patients had a 3.03 (1.0/0.33) times higher risk of dying of breast cancer compared to obese patients (p=0.002). These results indicate that non-hormonal mechanisms could be important.