An investigation is reported into the story comprehension abilities of four groups of children: those with typical specific language impairment (SLI-T), those with pragmatic language impairments who were not autistic (PLI), those with high-functioning autism (HFA) and typically developing controls. The story comprehension task required children to answer questions about the literal content of the story, as well as questions involving two types of inferences: text-connecting and gap-filling. The control children outscored the three clinical groups on story comprehension, but the group means of the clinical groups did not differ. However, categorical examination of the data revealed that children with pragmatic difficulties related to HFA were more likely to have specific inferencing deficits. Error analysis suggested that all children could make inferences, but these were not always relevant to the story context. This supports the notion of weak central coherence underlying deficits in inferencing. There were no group differences on story recall. However, there was a strong relationship between story comprehension and recall, in that those who had better comprehension tended to have better recall. It is concluded that comprehension aids recall by enabling the listener to build a more stable mental representation of the story. The pragmatic deficits seen in autism compromise this process.