Despite growing interest in improving cognitive abilities across the lifespan through training, the benefits of cognitive training are inconsistent. One powerful contributor may be that individuals arrive at interventions with different baseline levels of the cognitive skill being trained. Some evidence suggests poor performers benefit the most from cognitive training, showing compensation for their weak abilities, while other evidence suggests that high performers benefit most, experiencing a magnification of their abilities. Whether training leads to compensation or magnification effects may depend (...) upon the specific cognitive domain being trained and the training approach implemented. To clarify the association between individual differences in baseline cognitive ability and training gains as well as potential moderators, we conducted a systematic meta-analysis of the correlation between these two variables. We found evidence of a significant meta-correlation demonstrating a compensatory effect, a negative association between initial ability on a trained cognitive process and training gains. Too few papers met our search criteria across the levels of proposed moderators of cognitive domain and training approach to conduct a reliable investigation of their influence over the meta-analytic effect size. We discuss the implications of a compensatory meta-correlation, potential reasons for the paucity of qualifying papers, and important future directions for better understanding how cognitive trainings work and for whom. (shrink)