The platinum-based drugs cisplatin, carboplatin and oxaliplatin are often used for chemotherapy, but drug resistance is common. The prediction of resistance to these drugs via genomics is a challenging problem since hundreds of genes are involved. A possible alternative is to use mass spectrometry to determine the propensity for cells to form drug-DNA adducts?the pharmacodynamic drug-target complex for this class of drugs. In this paper, the feasibility of predictive diagnostic microdosing was assessed in non-small cell lung cancer (NSCLC) cell culture and a pilot clinical trial. Accelerator mass spectrometry (AMS) was used to quantify [<
sup>
14<
/sup>
C]carboplatin-DNA monoadduct levels in the cell lines induced by microdoses and therapeutic doses of carboplatin, followed by correlation with carboplatin IC<
sub>
50<
/sub>
values for each cell line. The adduct levels in cell culture experiments were linearly proportional to dose (R<
sup>
2<
/sup>
= 0.95, p <
0.0001) and correlated with IC<
sub>
50<
/sub>
across all cell lines for microdose and therapeutically relevant carboplatin concentrations (p = 0.02 and p = 0.01, respectively). A pilot microdosing clinical trial was conducted to define protocols and gather preliminary data. Plasma pharmacokinetics (PK) and [<
sup>
14<
/sup>
C]carboplatin-DNA adducts in white blood cells and tumor tissues from six NSCLC patients were quantified via AMS. The blood plasma half-life of [<
sup>
14<
/sup>
C]carboplatin administered as a microdose was consistent with the known PK of therapeutic dosing. The optimal [<
sup>
14<
/sup>
C]carboplatin formulation for the microdose was 10<
sup>
7<
/sup>
dpm/kg of body weight and 1% of the therapeutic dose for the total mass of carboplatin. No microdose-associated toxicity was observed in the patients. Finally, additional accruals are required to significantly correlate adduct levels with response.