Most neuropsychological evidence for category specific deficits in reading and face recognition comes from small-N studies of patients recruited on the basis of selective deficits. Taking a different approach, we recruited patients based on lesion location (Posterior Cerebral Artery-stroke) rather than symptomatology. 58 patients and 31 controls were included. We used a novel paradigm assessing recognition and memory for words, objects, and faces (the WOF-test), as well as typical tests of face recognition (CFMT) and reading aloud (RTs and word length effect). In a case series design, we analyse both group performance (left vs right hemisphere lesion), and single subject deficits, focusing on hemispheric differences in performance with faces and words. For the WOF-test and the CFMT, there were no significant differences in performance between lesion groups. Also, the proportion of patients in the left and right hemisphere groups with face recognition deficits and visual word processing deficits, respectively, did not differ. However, the severity of impairments in reading out loud was greater in the left hemisphere group. Four patients fulfilled the statistical criteria for a dissociation with impaired reading and preserved face recognition performance. These results suggest that face and word processing may be more bilaterally distributed than indicated by single case neuropsychological studies and functional imaging in neurotypicals. However, the observed dissociations indicate that word recognition may rely in part on (left) lateralized processes not involved in face recognition. This complements previous evidence that face recognition may be selectively impaired by right hemisphere stroke.