The influence of orthographic transparency on the prevalence of dyslexia sub-types was examined in a review of multiple-case studies conducted in languages differing in orthographic depth (English, French, and Spanish). Cross-language differences are found in the proportion of dissociated profiles as a function of the dependent variables (speed or accuracy), the classification method (classical vs. regression-based methods), and the control sample (chronological age vs. reading level controls). The classical method results in a majority of mixed profiles, whereas the regression-based method results in a majority of dissociated profiles. However, the regression-based method appears to result in less reliable subtypes within and between languages. Finally, reading-level comparisons revealed that the phonological subtype reflects a deviant developmental trajectory across all languages, whereas the surface subtype corresponds to a delayed developmental trajectory. The results also indicate that reading speed should be considered to correctly classify dyslexics 25 into subtypes, at least in transparent orthographies.