The core function of working memory is to store and process different information at the same time. It plays an important role in many complex cognitive activities, and also has an important impact on individual learning, reasoning, intellectual activities and problem solving. Working memory will be improved after being trained by different methods, such as training the individual's memory strategy, attentional controlled ability (including updating, switching and inhibition), and memory span training. Training on working memory tasks has been shown to improve performance on the same and related tasks and, in some cases, generalizes to other cognitive domains. Math skills rely on the basic knowledge of mathematical, as well as the development of general cognitive abilities. If working memory training can promote general cognitive ability, theoretically it can also transfer to improve mathematical ability. The research found that effect of working memory training can be transferred to math skills such as number sense, phonological space ability, and reasoning ability. Working memory training can also promote mathematical computing ability by improving phonological working memory and spatial ability; and it can promote complex mathematics processes such as problem representation, pattern recognition, problem solving, and strategies, by improving central executive system of working memory. However, the far transfer effects of working memory training on the individual's mathematical, fluid intelligence and other aspects are not consistent. Evidence showed that the immediate effect of far transfer is obvious, but the far transfer effect will be gradually weaken after three or six month. Therefore, further evidence of the effect of working memory training on the transfer of mathematical ability would be more clear if the cognitive processes of different mathematical tasks were distinguished. Different types of students with math learning difficulty have different features of cognitive functions, which would help to find the definite training effect of each training task. These findings suggest that the extent to which different math skills share neurocognitive systems and/or processes may predict the extent to which the effects of working memory training on one will transfer to the other. The evidence of neuroimaging may be another support of the improvement of mathematics ability by working memory training in the future.