Background: Autism exists across a wide spectrum and there is considerable debate as to whether children with Asperger's syndrome, who have normal language milestones, should be considered to comprise a subgroup distinct other from high-functioning children with autism (HFA), who have a history of delayed language development. Magnetic resonance imaging (MRI) studies of autism are in disagreement. One possible reason is that the diagnosis of autism takes precedence over Asperger's syndrome and a distinction in language acquisition is rarely made. We therefore planned to examine a whole brain hypothesis that the patterns of grey matter differences in Asperger's syndrome and HFA can be distinguished. Methods: We used voxel-based computational morphometry to map grey matter volume differences in 33 children with either Asperger's syndrome or high-functioning autism compared to 55 typical developing control children balanced for age, IQ, gender, maternal language and ethnicity. Results: Children with HFA had significantly smaller grey matter volumes in subcortical, posterior cingulate and precuneus regions than the Asperger's group. Compared to controls, children with HFA had smaller grey matter volumes in predominantly fronto-pallidal regions, while children with Asperger's had less grey matter in mainly bilateral caudate and left thalamus. In addition we found a significant negative correlation between the size of a grey matter cluster around BA44 language area and the age of acquisition of phrase speech in the children with HFA. When the groups were combined we confirmed a mixed picture of smaller grey matter volumes in frontal, basal ganglia, temporal and parietal regions. Conclusions: Our study suggests that the underlying neurobiology in HFA and Asperger's syndrome is at least partly discrete. Future studies should therefore consider the history of language acquisition as a valuable tool to refine investigation of aetiological factors and management options in pervasive developmental disorders.