Although arithmetic skills are crucial cognitive abilities, numeric competence impairments affect a significant portion of the young population. These problems produce a high socio-economic cost by negatively affecting scholastic and work performance. The parietal cortex is the brain area that is classically associated with numeric processing, but it is still debated whether other cortical areas are involved, and only a few studies tried to directly assess the causal link between brain and this cognitive function by using transcranial random noise stimulation, tRNS. This non-invasive electric stimulation device has been shown to enhance activity in the underlying cortex. We tested three groups of participants with equivalent arithmetic skills – an arithmetic ‘screening’ was administered. One group was stimulated by tRNS on the frontal lobe, another on the parietal lobe, and a third group was assigned to the placebo condition. During the stimulation, participants performed a subtraction verification task. To investigate long-term effects of tRNS, the task was repeated seven days later without stimulation. Aside previously-tested (familiar) subtractions, in the second experimental session unfamiliar subtractions were also administered. We found that, compared to placebo, parietal and frontal stimulation significantly reduced reaction times immediately, and enhanced accuracy after seven days. This benefit encompassed both familiar and unfamiliar subtractions. These results suggest that modulation of frontal and parietal cortices may ameliorate basic arithmetic skills by benefitting working memory function. This could open new avenues for neuro-restorative applications of brain stimulation.