Several studies have highlighted that children with reading comprehension difficulties also have problems in tasks that involve telling a story, in writing or verbally. The main differences identified regard poor comprehenders’ lower level of coherence in their productions by comparison with good comprehenders. Only one study has compared poor and good comprehenders’ performance in both modalities (oral and written), however, to see whether these modalities differently influence poor comprehenders’ performance. We qualitatively and quantitatively compared the performance of good and poor comprehenders in oral and written narrative tasks with the aim of shedding light on this issue. Regression analyses were also used to explore the role of working memory and vocabulary in explaining individual differences. Our results showed that the two groups produced narratives of comparable length, with similar percentages of spelling mistakes, whereas they differed in terms of the quality of their narratives, regardless of the modality. These differences were qualified by analyzing the children’s use of connective devices, and poor comprehenders were found to use a higher proportion of additive devices than good comprehenders. Regression analyses showed that working memory (particularly the intrusion errors measure) explained a modest part of the qualitative differences in narrative production. Implications for our theoretical understanding of poor comprehenders’ profiles and education are discussed.