We studied the impact on survival of changes in breast cancer patients' distribution by lymph node status at the time of diagnosis. Our study included breast cancer patients diagnosed from 1978 to 1994 in Denmark, where the treatment schemes for breast cancer patients were fairly stable, and where mammography screening was limited. We measured lymph node status by the proportion of positive lymph nodes of all excised lymph nodes, as assessed by a pathologist. This measure was available for two-thirds of the breast cancer patients. The outcome was 5-year relative survival. Changes in lymph node status distribution explained half of the improvement in 5-year relative survival, and seem to be the single most important cause behind the improved survival of breast cancer patients in Denmark.