The paper analyzes the relation among tumor size (T), estrogen receptor (ER) status, and labeling index (LI) and their relative merits in predicting the relapse-free (RFS) and overall survival (OS) in 215 node-negative women with primary breast cancer. All patients were subjected to Halsted or modified radical mastectomy; none received postoperative irradiation or systemic adjuvant therapy. The 5-year RFS was 75.3% and OS 89.0%. When singly tested, LI and ER were able to significantly predict RFS and OS. In contrast, T influenced only RFS but not OS. The multivariate analysis showed that, in the presence of the two other variables, only LI retained its prognostic significance both for time to relapse (p = 0.0044) and survival (p = 0.035). From the present findings, LI appears to be a new important prognostic variable in the selection of high risk patients for whom adjuvant systemic therapy should perhaps be part of their primary treatment.