We have shown that adjuvant cyclophosphamide (C), epirubicin (E), 5-FU (F) chemotherapy improves survival compared to CMF in pre-menopausal women with lymph node-positive (NP) breast cancer (NCIC-CTG MA5, JCO,1998). In that study, 5 cases of acute leukemia (AL) were seen among 359 patients receiving CEF, a crude rate of 1.4%. To refine the estimate of risk of AL after E-based chemotherapy, we used a competing risk method (Pepe and Mori, Stat Med, 1993) to calculate the marginal and conditional probabilities of secondary AL among 1463 pre-menopausal, NP women enrolled on 4 sequential NCIC-CTG adjuvant therapy trials (including MA5). Chemotherapy given included standard CEF, dose-intensive CE(F) + G-CSF given every 2 weeks, or standard-dose AC or CMF. Number of patients, follow-up, and number of observed cases of acute myeloid (AML) or acute lymphoblastic leukemia (ALL) according to treatment were as follows: CEE CE + G-CSF CMF AC No. of patients 521 97 633 212 Median F/U (yrs) 7.4 4.7 6.2 3.8 No. AML/ALL 5/2 0/0 1/0 2/0 Five patients with AL had AML with M4 or M5 morphology, and 3 had abnormalities of chromosome 11q23; one pt with AL6 1/2 years after CMF had a karyotype consistent with alkylator exposure (+8, -5, -19). Conditional probability of AL after 5 years according to adjuvant chemotherapy received: CEF: 1.8%; all E-containing regimens: 1.5%; non-E containing regimens (AC or CMF): 0.3%. After 5 years of follow-up, there was no apparent difference in conditional probability of AL between those receiving CEF (1.8%) vs AC (1.3%), but the number of patients who received AC is small and differences in follow-up duration are large. We conclude, in a larger group of patients with NP breast cancer and with longer follow-up, that the estimate of AL risk after E-based adjuvant chemotherapy has remained stable and, importantly, may not be different from currently used regimens such as AC or AC→T. These estimates are important in discussing treatment with patients and in the evaluation of future trials testing the addition of new agents to E-containing regimens in the adjuvant setting.