Working memory (WM) training has been proposed as a promising intervention to enhance cognitive abilities, but convincing evidence for transfer to untrained abilities is lacking. Prevalent limitations of WM training studies include the narrow assessment of both WM and cognitive abilities, the analysis of manifest variables subject to measurement error, and training dosages too low to likely cause changes in the cognitive system. To address these limitations, we conducted a 2-year longitudinal study to investigate the effects of working memory training on latent factors of working memory capacity, fluid intelligence and crystallized intelligence. One hundred twelve students initially attending 9th grade practiced a heterogenous set of validated WM tasks on a biweekly basis. A control group of 113 students initially attending 9th grade participated in the pretest and posttest. Broad and prototypical measures of fluid and crystallized intelligence served as measures of nearer and farer transfer. We found substantial and reliable training effects on the practiced WM tasks, as well as on a latent WM factor constituted by them. However, no transfer of training effects to fluid or crystallized intelligence were observed. These results add to the literature questioning the utility and validity of WM training as means of improving cognitive abilities.