The transfer effect of working memory training is to improve the performance of other tasks through working memory task training. Transfer is divided into near transfer and far transfer, where near transfer refers to training to improve the performance of tasks within the scope of working memory, and far transfer refers to training to improve the performance of tasks related to working memory. So, what is the mechanism of transfer effect? The current research on the transfer effect of working memory training presents contradictory results which have led to two mainstream views of transfer: the improvement of working memory ability or the improvement of working memory efficiency. The view of working memory ability improvement is that working memory training is similar to muscle training, in that it is thought that by continuously practicing a certain cognitive process, the overall resources associated with that process may be improved. The idea that working memory training has caused an increase in efficiency is that the transfer between tasks can be achieved by using strategies or increasing the degree of automation of working memory processing. The underlying assumption behind both views is that if an individual can improve the performance of all tasks that also require working memory through working memory training, then working memory training can be considered to improve the general working memory ability, otherwise, If working memory training can only improve the performance of tasks that are similar to the training task in processing or structure, it can only show that the training ed the individual to develop corresponding strategies, thereby improving the efficiency of working memory. Combined with most of the current research on working memory training, it can be found that transfer is limited to a certain extent, and the training of a single task will not improve all other tasks that require working memory, or even training of multiple working memory tasks, also failed to successfully transfer to other tasks. Therefore, most of the current research evidence supports the idea that transfer occurs due to improved working memory efficiency. A large number of recent studies have shown that the structural similarity between tasks has an important impact on transfer. Structural similarity refers to the sharing of processing and processing programs between tasks, which can be transferred to working memory training by developing similar strategies. This is beneficial to reduce the consumption of individual cognitive efforts and attention resources, thereby improving the individual's performance in training working memory tasks. But this strategy has a basic requirement for the application of untrained working memory tasks, that is, the training task and the untrained task have a similar structure. Therefore, the use of strategies may be an intermediary factor between structural similarity and the occurrence of transfer effects, and one way in which the use of strategy may play a role is that both tasks have similar sub-processes by forming specific strategies in similar sub-processing and information coding, cognitive burden is reduced, storage resources are released, and the performance of working memory task is promoted. Therefore, this article will focus on the influence of the similarity of task structure on the transfer effect of working memory training. First, it combs the recent research on the effectiveness and ineffectiveness of working memory training transfer. Second, it analyzes the reasons for the controversy from the perspective of task structure similarity. Finally, the future researches should pay more attention to the shortcomings in the previous research design, and study the mechanism of the working memory training effect of different groups separately.