The report discusses speech processes in reading by critically considering recent available evidence and by proposing a general model of speech processes. Stating that much work has been guided by the question of whether speech recoding precedes lexical access, the report proposes that a richer understanding of speech processes in reading must include attention to postlexical processes. Although there is little direct evidence, the report suggests that postlexical demands of reference securing make it reasonable to hypothesize that activation of some phonemic information is automatic and concurrent with lexical activation, and that the phonemic code is not a mere replica of a speech production. The report also proposes that experiments employing speech suppression paradigms are inadequate to detect the speech processes involved, and suggests both that there is a continuum of speech activation and that suppression operates at a higher level than the relevant speech processes. The report argues that recent research provides at least weak evidence for phonemic processes involving consonants, and also considers whether reading ability is related to the use of speech processes in reading. It is noted that there is considerable evidence to support this; however, there is little evidence presented to show that activation of speech codes fails to occur for less skilled readers. Evidence suggests that activation is slower or less automatic for less skilled readers, and that a speech code, once activated, deactivates more rapidly for such readers. The report concludes by suggesting that encoding conditions and memory demands may be related and that compensation between them can be arranged. (CRH)