The results of treatment for acquired dyslexia in a patient, P.S., are presented, cognitive analysis of the patient's performance provides evidence for proposing relatively selective impairments within the reading process to the orthographic input lexicon and to one or more components of sublexical processes for converting print to sound. Results of several contrasting treatment programmes for his deficits are reported, along with results of the same treatment approaches for patients with the same «diagnosis» (i.e. proposed locus of impairment in the cognitive processes underlying reading) or with different «diagnoses». The results illustrate that the relationship between the «diagnosis» and successful treatment is not simple. Rather, the reported results indicate the following departures from a 1:1 relationship between the locus of damage and the effective intervention: (1) contrasting treatment approaches can be equally appropriate for a given locus of damage, but the approaches may affect different aspects of language performance; (2) a given treatment strategy may be successful for some patients but not for others with the same putative level of damage (although this may be because the damage can take various forms); and (3) a given treatment strategy can be equally appropriate for several different levels of damage, perhaps in part because separate components of the treatment affect different levels of processing. These illustrations then serve as the basis for discussing the types of predictions that are possible regarding the effects of particular interventions on the basis of postulating a specific locus of disruption in the reading (or other cognitive) process in the patient to be treated.