Acute myeloid leukemia (AML) is a rare yet deadly cancer of the blood and bone marrow. Presently, induction chemotherapy with the DNA damaging drugs cytarabine (ARA-C) and idarubicin (IDA), known as 7 + 3, is the standard of care for most AML patients. However, 7 + 3 is a relatively ineffective therapy, particularly in older patients, and has serious therapy-related toxicities. Therefore, a diagnostic test to predict which patients will respond to 7 + 3 is a critical unmet medical need. We hypothesize that a threshold level of therapy-induced 7 + 3 drug-DNA adducts determines cytotoxicity and clinical response. We further hypothesize that in vitro exposure of AML cells to nontoxic diagnostic microdoses enables prediction of the ability of AML cells to achieve that threshold during treatment. Our test involves dosing cells with very low levels of 14C-labeled drug followed by DNA isolation and quantification of drug-DNA adducts via accelerator mass spectrometry. Here, we have shown proof of principle by correlating ARA-C- and DOX-DNA adduct levels with cellular IC50 values of paired sensitive and resistant cancer cell lines and AML cell lines. Moreover, we have completed a pilot retrospective trial of diagnostic microdosing for 10 viably cryopreserved primary AML samples and observed higher ARA-C- and DOX-DNA adducts in the 7 + 3 responders than nonresponders. These initial results suggest that diagnostic microdosing may be a feasible and useful test for predicting patient response to 7 + 3 induction chemotherapy, leading to improved outcomes for AML patients and reduced treatment-related morbidity and mortality.