Patients who develop therapy-related myelodysplasia/acute myeloid leukemia after autologous-hematopoietic stem cell (aHCT) transplant show lower expression levels of DNA repair genes in their pre-aHCT CD34+ cells. To investigate whether this leads to functional differences in DNA repair abilities measurable in patients, we adapted two plasmid-based host-cell reactivation assays for use in primary lymphocytes. Prior to applying these assays to patients who underwent aHCT, we wanted first to verify whether sample preparation affected repair measurements, as patient samples were simply depleted of erythrocytes (with hetastarch) prior to freezing, which is not the classical way to prepare lymphocytes prior to DNA repair experiments (with a density gradient). We show here that lymphocytes from healthy donors freshly prepared with hetastarch show systematically a higher level of double-strand break repair as compared to when prepared with a density gradient, but that most of this difference disappears after samples were frozen. Several observations points to granulocytes as the source for this effect of sample preparation on repair: 1) removal of granulocytes makes the effect disappear, 2) DSB repair measurements for the same individual correlate to the percentage of granulocytes in the sample and 3) nucleofection in presence of granulocytes increases the level of reactive oxygen species (ROS) in neighboring lymphocytes in a dosedependent manner (R2 of 0.95). These results indicate that co-purified granulocytes, possibly through the release of ROS at time of transfection, can lead to an enhanced repair in lymphocytes that obfuscates any evaluation of inter individual differences in repair as measured by host-cell reactivation. As a result, hetastarch-prepared samples are likely unsuitable for the assessment of DSB repair in primary cells with that type of assay. Granulocyte contamination that exists after a density gradient preparation, although much more limited, could have similar effects, but might be circumvented by freezing cells prior to analysis. © 2014 Lacoste et al.