Working memory (WM) training paired with transcranial direct current stimulation (tDCS) can improve executive function in older adults. The unclear mechanism of tDCS likely depends on tDCS intensity, and task relevant genetic factors (e.g., for WM: COMT val158met, DAT, BDNF val66met). Higher tDCS intensity does not always lead to greater cognitive gains, and genetic polymorphisms may modulate tDCS-linked WM improvements. To evaluate these factors, 137 healthy older adults provided DNA samples and received Visual and Spatial WM training paired with tDCS (sham, 1, 1.5, 2 mA). After one session of tDCS, significant group differences in WM performance were predicted by COMT val158met status. One month after training, there was a significant interaction of tDCS intensity, COMT genotype, and WM task. Specifically, val/val homozygotes benefited most from 1.5 mA tDCS on Visual WM and from 1 mA tDCS on Spatial WM. For met/met homozygotes, 2 mA resulted in significantly poorer performance compared to 1.5 mA on Spatial WM. While this pattern was observed with relatively small sample sizes, these data indicate that variations in COMT val158met may predict the nature of WM improvement after initial and longitudinal tDCS. This contributes to our understanding of the underlying mechanism by which tDCS affects behaviour.