Abstract Central coherence theory (Frith, U., 1989. Autism: Explaining the Enigma. Blackwell, Oxford.) is addressed by exploring linguistic processing in normally intelligent adults with either autism or Asperger syndrome, to test whether local coherence is impaired. Local coherence is the ability to make contextually meaningful connections between linguistic information in short-term or working memory. Experiment 1 demonstrated that individuals with an autism spectrum condition were less likely to use the sentence context spontaneously to provide the context-appropriate pronunciation of a homograph. Experiment 2 presented scenarios which had a situation and outcome which only cohered if a bridging inference was drawn. The clinical groups were less likely to select the most coherent (bridging) inference from competing alternatives. Experiment 3 demonstrated that individuals with an autism spectrum condition were less able to use context to interpret an auditorily presented ambiguous sentence. The findings from Experiments 2 and 3 suggest that individuals with autism or Asperger syndrome have a difficulty in achieving local coherence, while the evidence from Experiment 1 suggests a preference not to strive for coherence. Taken together, these results suggest that individuals with an autism spectrum condition are impaired in achieving local coherence, and they have a preference not to strive for coherence unless instructed to do so, or unless they make a conscious decision to do so. Moreover, the three experiments correlate with one another, which suggests that central coherence may be a unitary force in these different tasks. Of the two clinical groups, the autism group had the greater difficulty in achieving coherence. Possible explanations for the clinical groups' difficulty are explored.