Purpose: We previously reported evidence of true double dissociation between reading accuracy and reading rate in a large unselected sample of Hebrew-speaking fourth graders and a large clinical sample of adult Hebrew-speakers with dyslexia. The present study aimed to replicate and extend these findings to Arabic, which is structurally similar to Hebrew but has distinct linguistic and orthographic features. Method and results: In a nationally representative 4th grade sample (N = 236), we show that (1) around one third of children with dyslexia had impaired reading rate but intact accuracy whereas another third had impaired accuracy but intact rate, (2) there was a double dissociation with respect to additional (validation) measures of reading accuracy and rate (pseudowords and text), and (3) the accuracy-only and rate-only disability subtypes displayed distinct and non-overlapping cognitive-linguistic profiles. Conclusion: This evidence converges on the conclusion that accuracy-only and rate-only dyslexic subtypes represent true or "hard" subtypes in an absolute and not merely relative sense. We also found that the accuracy-only subgroup represents a group with broad language weaknesses, primarily phonological but also non-phonological. Finally, we discuss the resemblance between the present rate-accuracy typology and Wolf and Bowers' double-deficit typology.