The phonological-core variable-orthographic differences (PCVOD) model [van der Leij, & Morfidi (2006). Journal of Learning Disabilities, 39,[74][75][76][77][78][79][80][81][82][83][84][85][86][87][88][89][90] has been proposed as an explanation for the heterogeneity among dyslexic readers in their profiles of reading-related subskills. The predictions of this model were investigated in a sample of 72 Dutch secondary school students (dyslexics and controls). First, the PCVOD assumption was confirmed that phonological processing and orthographic competence are independent contributors to the prediction of reading fluency and spelling. Among the phonological processing tasks, phonological recoding explained substantial unique variance, but not phonemic awareness or rapid serial naming. Next, the dyslexic readers were divided into two subgroups based on high (ORTH + ) and low levels (ORTH -) of orthographic competence. Both subgroups performed below controls on all measures tapping phonological processing, reading and spelling but the ORTH + group performed as well as nondisabled controls on Dutch and English orthographic choice. As predicted by the model, there were no differences between the subgroups on the tasks that depend on phonological processing, with or without reading. There were differences on Dutch word reading fluency and spelling. Furthermore, the ORTH + subgroup outperformed ORTH -on tasks demanding speeded word processing such as 'flashed' presentation. This finding was independent of lexicality (words or pseudowords), language (Dutch or English) or response mode (lexical decision or typing), but restricted to silent reading. This supports the view that the ORTH + subgroup is better at identifying larger orthographic units. There was no indication of differences between the subgroups in reading experience. Our data, therefore, support the PCVOD model.