Background: In 1984, John Marshall made the case that one can use a model of the skilled reading system not only to interpret the acquired dyslexias, but also to interpret the developmental dyslexias, and the particular model of the skilled reading system he favoured for this purpose was the dual-route model. This claim has been a controversial one, with many researchers claiming that static models of adult skilled reading, such as the dual-route model, are inappropriate for understanding the process of reading development and, consequently, the developmental reading disorders. Aims: In this paper, we examine how Marshall's conjecture has fared over the past 20 years. Main Contribution: We evaluate Marshall's conjecture by examining evidence for developmental analogues of acquired surface and phonological dyslexia, by reporting new data on cases of "pure" developmental surface and phonological dyslexia, and by assessing the success of dual route versus connectionist accounts of these subtypes. We also report evidence that the dual-route model of skilled reading provides an accurate account of the reading performance of children at all stages of reading development. Conclusion: We conclude that Marshall's controversial claim has been vindicated by subsequent research.