A group of 21 participants with specific reading disorders was treated with a method derived from dual-route models and another group of 23 with a method derived from single-route models of reading. Both treatments were compared with four control treatments. The reading performance of each participant was compared with that of his or her chronological-age controls on the following variables: speed and accuracy of reading passage, isolated words and nonwords, and accuracy in homophone recognition. The treatment deriving from dual-route models produced significant improvements in the homophone recognition, compared to all other treatments. The treatment deriving from single-route models produced significant improvements, compared to all other treatments, in speed of word reading. Furthermore, these two treatments produced significant improvement with respect to all other treatments but one, in speed of nonword reading. These findings support the hypothesis that treatments derived from specific models of reading development are superior to other treatments. However the benefits obtained on the reading of isolated stimuli (words, homophones) did not significantly improve the reading of a passage. This fact suggests that treatments should include exercises involving passages or sentences.