The primary purpose of this study was to evaluate the prognostic significance of Ki-67, a cell proliferation-associated antigen, in a large group (n = 674) of axillary node-negative breast cancer cases with long-term follow-up and to correlate Ki-67 antigen expression with S-phase fraction. Ki-67 immunostaining was assessed both semiquantitatively and quantitatively. The statistical analysis focused on agreement between methods of Ki-67 quantification, agreement between Ki-67 and S-phase fraction, associations between Ki-67 and other clinical variables, and prognostic value of Ki-67. There was excellent agreement between the two methods of Ki-67 assessment (Spearman rank correlation, rsp = 0.91; P = 0.0001; n = 674) but only weak correlation between either semiquantitative or quantitative Ki-67 and S-phase fraction (rsp = 0.12 and rsp = 0.15, respectively). Ki-67 (overall median, 2%) was independent of tumor size and modestly related to other measures of tumor aggressiveness. Using a cutpoint of 5% (percentage of tumor cells), cases with high Ki-67 exhibited a significantly shorter disease-free survival (Padj = 0.004). In multivariate analysis, high Ki-67 was associated with a 1.8-fold increased risk of recurrence (P = 0.001). In the subgroup with S-phase data, the adjusted relative risk (hazard ratio, 1.9; P = 0.02) was unchanged by inclusion of S phase in the model. This suggests that Ki-67 provides significant independent prognostic information in addition to that contained in tumor size and S-phase fraction.