We examined the responsiveness to a 12‐week phonics intervention in 54 s‐grade Dutch children with dyslexia, and compared their reading and spelling gains to a control group of 61 typical readers. The intervention aimed to train grapheme–phoneme correspondences (GPCs), and word reading and spelling by using phonics instruction. We examined the accuracy and efficiency of grapheme–phoneme correspondences, decoding words and pseudowords, as well as the accuracy of spelling words before and after the intervention. Moreover, responsiveness to intervention was examined by studying to what extent scores at posttest could directly or indirectly be predicted from precursor measures. Results showed that the children with dyslexia were significantly behind in all reading and spelling measures at pretest. During the intervention, the children with dyslexia made more progress on GPC, (pseudo)word decoding accuracy and efficiency, and spelling accuracy than the typical reading group. Furthermore, we found a direct effect of the precursor measures rapid automatized naming, verbal working memory and phoneme deletion on the dyslexic children's progress in GPC speed, and indirect effects of rapid automatized naming and phoneme deletion on word and pseudoword efficiency and word decoding accuracy via the scores at pretest. Copyright © 2016 John Wiley & Sons, Ltd.