In working memory training studies, individual trajectories are known to vary considerably between participants. A better understanding of how individual differences affect training outcomes is important because it might inform the development of more effective training interventions. This study explored how measures of working memory, intelligence, sustained attention, training motivation, mindset, psychological well-being, perceived stress, and sleep quality affect initial training performance and rate of change. A total of 217 upper secondary students completed 12 weeks of adaptive dual-n-back in a classroom setting. We analyzed their self-reported training data using latent growth curve modeling. We found that working memory and intelligence predicted both, initial training performance and rate of performance change. Sustained attention and sleep quality predicted initial performance, but not the rate of change. Furthermore, we observed that participants who completed the intervention scored significantly higher on measures of working memory and intelligence and reported lower levels of perceived stress and higher levels of sleep quality at baseline compared to dropouts. In general, our study supports the magnification account with higher ability individuals starting out at a higher performance level and showing a higher rate of performance change, and moreover, being more likely to adhere to the training protocol.