This study compared the effects of a combined intervention of working memory (WM) and arithmetic reasoning (AR) vs WM intervention alone on the arithmetic reasoning performance of students with ADHD. Third- and fourth-grade elementary school students (n = 46) completed measures for AR, mathematical calculations, and WM. Participants were randomised using a minimisation approach taking age and IQ as variables of interest and assigned to one of the two groups: Combined Intervention (CI, n = 24) and Working Memory Intervention (WMI, n = 22). The results using GEE analysis indicated a significant group × time interaction (WaldX[superscript 2] = 6.414; gl = 2; p = 0.04) in AR performance in the immediate post-test. CI students showed significantly better performance on AR than WMI students immediately after intervention (pB = 0.042). There was an effect of time on mathematical calculations in the post-test (WaldX[superscript 2] = 48.305; gl = 2; p < 0.001). Despite the fact that the results for AR were not maintained in the deferred post-test, a combined intervention of WM and AR seems to be more efficient in improving arithmetic reasoning in ADHD students than a WM intervention alone. Nevertheless, this is not the case with other mathematical issues, such as calculation, where there was no significant difference between groups, but the effects had been maintained in the deferred post-test for both.