A single case study is reported of the dyslexic and only mildly aphasic pat
ient AT who almost exclusively produced visually related errors in reading
aloud. His performance in the auditory modality (auditory comprehension, ph
oneme discrimination) was nearly perfect, his repetition and oral naming ab
ilities for single-word production were almost unimpaired. Closer examinati
ons of his reading disturbance revealed a deficit in visual processing of o
rthographic materials which extended to Arabic numbers and other nonlinguis
tic symbols. Contrasting with this prelexical deficit, AT's performance in
reading aloud and also in visual lexical decision was strongly influenced b
y factor imageability, which could not be attributed to a general semantic
deficit. It is argued that AT's pattern of reading performance (prelexical
visual impairment and imageability effects on his visual errors) cannot be
accounted for by a discrete stage model of lexical-orthographic processing
allowing only one lexical candidate to be selected. Rather, the case AT pro
vides empirical evidence for cascaded processing between the lexical and se
mantic level. The cascade account suggests that not only one but several vi
sually related lexical candidates activate their corresponding semantic rep
resentations with different degrees of imageability.