The effectiveness of working memory (WM) training programmes is still a subject of debate. Previous reviews were heterogeneous with regard to participant characteristics of the studies included. To examine whether these programmes are of added value for children with learning disabilities (LDs), a systematic meta-analytic review was undertaken focusing specifically on LDs. Thirteen randomised controlled studies were included, with a total of 307 participants (age range = 5.5–17, Mean age across studies = 10.61, SD = 1.77). Potential moderator variables were examined, i.e., age, type of LD, training programme, training dose, design type, and type of control group. The meta-analysis indicated reliable short-term improvements in verbal WM, visuo-spatial WM, and word decoding in children with LDs after training (effect sizes ranged between 0.36 and 0.63), when compared to the untrained control group. These improvements sustained over time for up to eight months. Furthermore, children > 10 years seemed to benefit more in terms of verbal WM than younger children, both immediately after training as well as in the long-term. Other moderator variables did not have an effect on treatment efficacy.