The purpose of this study was to examine the efficacy of a method to improve reading fluency in children with dyslexia. The method, which we named "subsyllabic," was aimed at automatizing the recognition of syllables within words in connected texts, presented by ad hoc software. Two versions of this method--one self-paced and the other one with automatic syllable identification-were compared to a method based on phonemic awareness, assisted reading, and other psycholinguistic exercises. The efficacy of the two versions of the subsyllabic method was further studied by repeating the first version twice and the second version three times using an AB design, with each phase lasting approximately 3 months. This part of the study provided not only follow-up data but also useful information on if and how fluency may change after repeated treatment. Outcomes obtained by a total of 63 children with dyslexia suggested that the subsyllabic method was superior to the control method and that the use of an automatic presentation of target syllables produced better results. Furthermore, we observed that fluency improved approximately at the same rate after each treatment repetition. Our data support the possibility of improving reading fluency at a significant clinical level, at least for regular orthographies. The crucial component of the subsyllabic method seems to be the facilitation of syllable recognition within words in connected texts and the emphasis on their rapid recognition using an automatized procedure.