Evidence from a large number of laboratories indicating a transient/magnocellular deficit in dyslexics is cited as a background to addressing the question of the significance of such a visual processing deficit in learning to read. This question is considered in terms of possible direct contributions to difficulty in learning to read based on multiple regression studies and in terms of the reading errors that would be expected if a transient/magnocellular deficit hindered the ability to learn to read. Evidence supporting both positions is discussed. Finally, it is suggested that a complete model of dyslexia may have to account for both the reading of single words (which most current models address) and the reading of continuous text (which most current models do not address).