In this study, I sought to investigate whether visuospatial working memory (VWM) explains individual differences in reading fluency growth. One hundred and thirty Japanese junior and senior high school students were administered the Mr. Peanut and reversed Corsi tasks, respectively as measures of VWM, followed by three waves of reading fluency assessment. To assess reading fluency, a novel reading fluency measure was constructed by combining the participants' comprehension scores with their reading speeds into a single measure using many-facet Rasch measurement. Latent growth curve analyses revealed that VWM significantly predicted variation in the individual growth trajectories at the initial status but did not significantly predict rate of change. Results of Rasch analyses indicated that the novel reading comprehension-speed measure assessed a unidimensional construct, thus suggesting that the construction of this measure is both a practical and valid approach to estimating reading fluency as well as reading passage difficulty.