It is widely accepted that developmental dyslexia results from some sort of phonological deficit. Yet, it can be argued that phonological representations and their processing have been insufficiently tested in dyslexia research. Firstly, claims about how tasks tap into certain kinds of representations or processes are best appreciated in the light of an explicit information-processing model. Here, a cognitive model of lexical access is described, incorporating speech perception, reading and object recognition. The model emphasises that phonological forms of lexical items are distinct from non-lexical phonological representations Secondly, phonology, as a linguistic discipline, teaches us that there is much more to it than phonemic categorisation and awareness. The phonological level of representation also embodies phonotactic regularities, patterns of phoneme assimilation and alternation, as well as supra-segmental knowledge pertaining to syllable structure, stress, intonation and rhythm. All those aspects are in part language-dependent, and therefore must be learnt by children in order to become proficient native speakers and listeners. If all levels of phonological representations were affected in dyslexia, dyslexic children would presumably have difficulties acquiring these aspects of their language. This prediction is as yet untested. A possible research agenda is outlined, aiming to provide a more comprehensive assessment of the phonological theory of dyslexia.