The purpose of this study was to determine factors associated with lymph node metastasis among hormonally-responsive breast cancer patients > or = 70 years old, and to develop and validate a clinical prediction rule to predict the risk of lymph node metastasis in this population. Nodal evaluation in elderly women with breast cancer remains controversial. The ability to predict which elderly patients may be node-negative may spare them the morbidity of lymph node evaluation. Hormone-receptor positive breast cancer patients > or = 70 years old who participated in a prospective multicenter trial were divided into a training set (n = 554) and a test set (n = 146). Univariate and multivariate analyses were conducted to determine factors predictive of final nodal status. A clinical prediction rule was developed on the training set, and validated in the independent test set. Median patient age was 76; median tumor size was 1.4 cm. 15.9% and 16.2% were LN+ in the training and test sets, respectively. On univariate analysis, patient age, tumor size, palpability, grade, and lymphovascular invasion predicted lymph node status. On multivariate analysis, patient age, tumor size, and lymphovascular invasion remained significant. A prediction rule was created; patients were categorized into quartiles by predicted risk. 5.4% and 0% of patients in the lowest quartile were node positive in the training and test sets, respectively. Some elderly breast cancer patients at low likelihood of lymph node metastasis may be spared lymph node evaluation.