This paper reviews evidence in support of the phonological deficit hypothesis of dyslexia. Findings from two experimental studies suggest that the phonological deficits of dyslexic children and adults cannot be explained in terms of impairments in low-level auditory mechanisms, but reflect higher-level language weaknesses. A study of individual differences in the pattern of reading skills in dyslexic children rejects the notion of 'sub-types'. Instead, the findings suggest that the variation seen in reading processes can be accounted for by differences in the severity of individual children's phonological deficits, modified by compensatory factors including visual memory, perceptual speed and print exposure. Children at genetic risk who go on to be dyslexic come to the task of reading with poorly specified phonological representations in the context of a more general delay in oral language development. Their prognosis (and that of their unaffected siblings) depends upon the balance of strengths and difficulties they show, with better language skills being a protective factor. Taken together, these findings suggest that current challenges to the phonological deficit theory can be met.