The double-deficit hypothesis (DDH) of the developmental dyslexias (Wolf and Bowers, "Journal of Educational Psychology", 91, 415-438, 1999) was investigated with 149 adolescents and young adults (age range?=?16 to 24 years) with dyslexia. Confirmatory factor analysis indicated that a two-factor model with separate naming speed (NS) and phonological awareness (PA) constructs was superior to a one-factor model, supporting the assumption within the DDH that NS is a source of reading dysfunction separable from PA. Hierarchical multiple regression analyses resulted in findings that were only partially supportive of the DDH. NS was predictive of word reading, spelling, and reading fluency beyond PA and verbal intellectual ability, but not pseudoword reading and timed and untimed reading comprehension. Examination of DDH subtypes did not support the core assumption of the DDH that the double-deficit subtype would have more impaired reading skills than both of the single-deficit subtypes. The NS deficit subtype was found to be more prevalent than the double-deficit and PA deficit subtypes within the subgroup of dyslexics with impairment in reading fluency. Overall results provided mixed support for the DDH and pointed to the need for the inclusion of additional abilities within theories of the underlying mechanisms disrupted in dyslexia.