In order to assess the efficacy of phonological versus surface dyslexia subtyping within an adult dyslexic population, 45 adult dyslexics were assessed on the Castles and Coltheart (1993) irregular word and non-word reading tasks. Based on the performance of a matched group of 28 non-dyslexics, between 62 and 75% of the dyslexics were divided into phonological and surface subtypes. Phonological dyslexics were those individuals who presented evidence of relatively poor or inefficient non-word reading, whereas the surface dyslexics were those who indicated relative difficulties with irregular word reading. The proportions of dyslexics within each subtype varied according to the use of accuracy-based or latency-based procedures, but were consistent with previous findings with children. Subsequent comparisons between the groups on measures of phonological processing, lexical access and word knowledge/recognition indicated few differences between the two subtypes and, in particular, no differences on measures of phonological ability. The lack of observed differences on such measures casts doubt on the efficacy of this procedure for explaining individual differences amongst adult dyslexics and as a practical procedure for diagnosis and intervention.