•Five individuals with acquired dyslexia reported excellent word reading skills.•The same individuals had difficulty reading nonwords.•But nonword reading drastically improved when the nonwords contained morphemes.•The data point to a pre-lexical activation of morphemes during reading.•Lexical, phonological, and morphological processing were clearly dissociable. The current study investigated the influence of morphological structure in nonword reading in a case series of individuals with acquired dyslexia following brain damage. The aim of the study was to test the separate influence of embedded stems and suffixes on reading skills by comparing four different types of complex nonwords: stem + suffix (e.g., nightness), stem + non-suffix (e.g., nightlude), non-stem + suffix (e.g., nishtness), and non-stem + non-suffix (e.g., nishtlude); and two types of words: suffixed (e.g., baker) and non-suffixed (e.g., diamond). We report five individuals with excellent word reading skills, including no difficulties in reading morphologically complex words, but who had difficulty in reading nonwords. Nonword reading skills drastically improved when the nonwords were composed of a real stem compared to a non-stem, or a real suffix compared to a non-suffix. In one of the individuals (RF), a significant stem-by-suffix interaction was observed, suggesting that they additionally benefited when letter-strings were decomposable into two morphemes. Another individual’s reading (SH) was facilitated by the presence of suffixes, but not by the presence stems. The impact of morphological structure on nonword reading, clearly observed in all five individuals, points to a pre-lexical activation of morphemes during reading. The results suggest that complex word reading involves three dissociable mechanisms: whole word processing, phonological decoding, and pre-lexical morpheme activation.