Purpose: Cytarabine, a key chemotherapy agent for acute myeloid leukemia (AML) treatment, is deaminated into inactive uracil-arabinoside by cytidine deaminase. This deamination leads to samples stability issues with respect to clinical pharmacokinetic trials. The aim of our study was to study in vitro cytarabine stability in blood samples obtained from AML patients. Methods: Cytarabine quantification was performed using a fully validated LC/MS/MS method. In vitro cytarabine stability was assessed at room temperature over 24 h in samples coming from 14 AML patients and 7 control patients (CTRL) with no hematological malignancy. In vitro concentrations versus time data were analyzed using a noncompartmental approach. Results: Cytarabine in vitro area under the curve (AUCIVlast) was 22-fold higher in AML samples as compared to CTRL samples (AML mean (standard deviation (SD)), 51,829 (27,004) h ng/mL; CTRL mean (SD), 2356 (1250) h ng/mL, p = 0.00019). This increase was associated with a prolonged in vitro degradation half-life (t1/2IVdeg AML mean (SD), 15 (11.8) h; CTRL mean (SD), 0.36 (0.37) h, p = 0.0033). Multiple linear regression analysis showed that AML diagnosis significantly influenced t1/2IVdeg and AUCIVlas relationship. Conclusion: Cytarabine stability is higher in AML than in CTRL samples. The absence of correlation between t1/2IVdeg and AUCIVlast in AML samples suggests that in vitro cytarabine degradation in AML is complex. These results open perspectives including the evaluation of the clinical relevance and the involved molecular mechanisms.