A previous Phase I study (Seiter K et al, 2002) had shown antileukemic activity of temozolomide (TMZ) in acute myeloid leukemia (AML). We conducted a Phase II study of TMZ in poor risk patients aged 60 years and over with AML. Patients were eligible if they had received no prior inductions and had one of the following: poor risk cytogenetics (complex abnormalities, deletions of chromsome 7 or 5, 11q23 abnormalities, inv) or secondary leukemia (prior hematologic disorder or therapy-related leukemia). Patients who had not achieved remission or had relapsed after one prior induction course, regardless of other risk factors, were also eligible. Of 43 evaluable patients, 17 had received one prior induction and 26 had received no prior therapy - of the latter group, 12 had secondary leukemia and the remaining 14 had poor risk cytogenetics. The dose of TMZ was 200 mg/m2 orally daily for 7 days. Patients achieving partial response (PR) were eligible to receive a second identical course. Patients achieving complete response (CR or CRp) were eligible to receive TMZ 200 mg/m2 orally daily for 5/28 days for up to 6 postremission cycles. The treatment was well-tolerated, with mild to moderate myelosuppression and minimal nausea. Most patients received therapy on an outpatient basis. Four patients achieved complete response (3 CR and 1 CRp), 2 achieved PR and 2 had marrow aplasia at Day 30–35; the remaining 35 patients did not respond. Of the 4 patients achieving CR, one relapsed after 3 months, while the remaining patients remain in CR after 7, 12 and 29 months respectively. The 2 PR patients did not respond further to a second induction course. Because sensitivity to TMZ has been previously linked to expression of the DNA repair enzyme O6-methylguanine-DNA methyltransferase (MGMT), we assessed baseline levels of MGMT in leukemia blasts pre-treatment by Western blot. Of 24 evaluable samples, 10 had no detectable MGMT; this group included all 4 CR patients, one PR and one aplasia. Of the 14 with detectable MGMT there were no antileukemic responses. The remaining samples were either not available or not evaluable for MGMT. In conclusion, TMZ therapy was well tolerated and had antileukemic activity with approximately a 10% CR rate in older poor risk AML patients. MGMT appears to be predictive of response - all of the responders had undetectable MGMT activity, while no antileukemic activity was seen in MGMT positive cases. This assay may therefore be useful in selecting potential candidates for this treatment. Confirmation of this finding in a larger series is warranted. Furthermore, pharmacologic reduction in MGMT levels could potentially increase response rates to TMZ and warrants study.