A search for prognostic discriminants of treatment failure in the tenth postmastectomy year was undertaken in 614 patients enrolled in protocol no. 4 of the National Surgical Adjuvant Project for Breast Cancers treated by radical mastectomy. Exploratory analyses of 38 pathologic and 6 clinical features disclosed 16 and 13 variables significantly related to nodal status and treatment failure, respectively. However, multivariate analyses with life tables adjusted or controlled for nodal status revealed that patients whose tumors measured less than 2 cm had a more favorable clinical course. All of the characteristics were also explored when patients were stratified according to numbers of nodal metastases, the most significant prognostic discriminant for disease-free survival in the tenth year. A germinal center predominance lymph node pattern and high histologic grade according to our conventional grading method were observed to adversely influence disease-free survival in patients with negative nodes. A strong trend in this manner was also observed with tumors greater than or equal to 2 cm and types 3 and 5 scar cancers. Histologic grade and tumor size were significantly recognized as discriminatory in patients with 4+ nodes but no factors were significantly observed in the group of patients with 1 to 3 positive nodes. A possible explanation for this inconsistency is discussed. For the most part, these factors were also encountered by us previously at the 5-year postoperative period. Only the presence of 13+ nodal metastases and types 3 and 5 scar cancers and cancers without scar appeared to discriminate success or failure between the fifth and tenth year. It is therefore concluded that nodal category, germinal center predominance, histologic grade, and tumor size, in the contexts noted, represent strong prognostic discriminants exerting a rather constant influence on disease-free survival at least to the tenth postmastectomy period. The prognostic value of categorizing those patients with 4+ positive nodes into subgroups with 4 to 6, 7 to 12 and 13+ is reaffirmed. The actual disease-free survival of patients in the tenth year was not strikingly different from that observed previously by us at 5 years in this cohort.