Between 1972 and 1974, patients were entered into a National Surgical Adjuvant Breast and Bowel Project (NSABP) trial to evaluate L-phenylalanine mustard (L-PAM) as an adjuvant to mastectomy in patients with primary breast cancer and pathologically positive axillary nodes. Overall, findings through 10 years of observation indicate an 8% difference in disease-free survival (DFS) (P = .06) and a 5% difference in survival (P = .4). Women less than or equal to 49 years of age who received L-PAM demonstrate a significant (P = .03) prolongation of DFS and a significant (P = .05) survival benefit compared with those who received a placebo. There is a 37% reduction in their mortality and a cumulative odds of survival of 1.67. In that age group, both those with one to three and greater than or equal to 4 positive nodes benefited, but the advantage was greater when there were fewer positive nodes. There was a significant (P = .009) reduction in mortality (64%) and a cumulative odds of survival of 3.25 in patients less than or equal to 49 years old with one to three positive nodes. No advantage from L-PAM was observed in patients greater than or equal to 50 years old. When L-PAM response is related to nuclear grade, a marker of tumor differentiation, there is a highly significant improvement in DFS (less than .001), distant DFS (.001), and survival (.004) through 10 years of observation for all patients with tumors classified as nuclear grade poor (poorly differentiated), regardless of age and nodal status. Mortality was reduced by 32% and the cumulative odds of survival was 2.10 at 10 years. Of singular importance is the observation of a benefit in those greater than or equal to 50 years of age as well as in those less than or equal to 49 years of age with poorly differentiated tumors. There is a significant (P = .003) survival benefit for those in the older age group with a 37% reduction in mortality, and a cumulative odds of survival of 2.75. Patients in both age groups with well-differentiated tumors demonstrated no benefit from L-PAM. Those who are older seem to do less well following chemotherapy. We conclude that tumor differentiation is a better discriminant of response to chemotherapy than is age, and that both younger and older women are responsive to adjuvant chemotherapy when they have poorly differentiated tumors.