Can cognitive abilities such as reasoning be improved through working memory training? This question is still highly controversial, with prior studies providing contradictory findings. The lack of theory-driven, systematic approaches and (occasionally serious) methodological shortcomings complicates this debate even more. This review suggests two general mechanisms mediating transfer effects that are (or are not) observed after working memory training: enhanced working memory capacity, enabling people to hold more items in working memory than before training, or enhanced efficiency using the working memory capacity available (e.g., using chunking strategies to remember more items correctly). We then highlight multiple factors that could influence these mechanisms of transfer and thus the success of training interventions. These factors include (1) the nature of the training regime (i.e., intensity, duration, and adaptivity of the training tasks) and, with it, the magnitude of improvements during training, and (2) individual differences in age, cognitive abilities, biological factors, and motivational and personality factors. Finally, we summarize the findings revealed by existing training studies for each of these factors, and thereby present a roadmap for accumulating further empirical evidence regarding the efficacy of working memory training in a systematic way.