A series of 743 consecutive cases of operable breast cancer, admitted and treated at the Istituto Nazionale Tumori of Milan from 1969 to 1970, was analyzed by a multivariate statistical method to evaluate a) the variables of the host and the primary tumor associated with the frequency of nodal metastases, b) the variables that significantly affect survival, and c) the identification of homogeneous risk groups. As regards the frequency of regional node metastases, they were more frequently observed in young than in old patients with large tumors (P values 10(-5) and 3 X 10(-5), respectively). Tumors that originated in the axillary tail, upper, outer and central quadrants were significantly associated with a higher rate of node metastases (P = 0.002). Each of these variables maintained its significant value when adjusted by the other two. Survival was affected at a statistically significant level by the age of the patients (P = 2 X 10(-4) ), the pathologic diameter of the primary tumor (P less than 10(-6) ), and the number of metastatic regional nodes (P less than 10(-6) ). The number of involved nodes appears to be the most relevant factor in the assessment of prognosis of patients with positive nodes, Age of the patients, size of the primary tumor, and number of involved nodes maintain their own statistical significance when each is adjusted by the remaining two. The site of origin of the primary tumor, even if associated with the frequency of regional node metastases, did not affect survival. Three groups with a significantly different risk of death were identified in patients with negative lymph nodes and three groups in patients with positive nodes. It is concluded that age, size of the primary, and number of involved lymph nodes are important pieces of information that clinicians should have at hand following radical surgery, not only to make a prognosis, but also to identify groups of patients with high risk of death on which the role of adjuvant treatment should be evaluated.