Frontotemporal lobar degeneration (FTLD) is clinically, pathologically and genetically heterogeneous. Recent descriptions of a pathological sub-type that is ubiquitin positive, TDP-43 negative and immunostains positive for the Fused in Sarcoma protein (FUS) raises the question whether it is associated with a distinct clinical phenotype identifiable on clinical grounds, and whether mutations in the Fused in Sarcoma gene (FUS) might also be associated with FTLD. Examination of a pathological series of 118 cases of FTLD from two centres, showing tau-negative, ubiquitin-positive pathology, revealed FUS pathology in five patients, four classified as atypical FTLD with ubiquitin inclusions (aFTLD-U), and one as neuronal intermediate filament inclusion disease (NIFID). The aFTLD-U cases had youthful onset (22-46 years), an absence of strong family history, a behavioural syndrome consistent with frontotemporal dementia (FTD) and severe caudate atrophy. Their cognitive/behavioural profile was distinct, characterised by prominent obsessionality, repetitive behaviours and rituals, social withdrawal and lack of engagement, hyperorality with pica, and marked stimulus-bound behaviour including utilisation behaviour. They conformed to the rare behavioural sub-type of FTD identified previously by us as the "stereotypic" form, and linked to striatal pathology. Cognitive evaluation revealed executive deficits in keeping with subcortical-frontal dysfunction, but no cortical deficits in language, perceptuospatial skills or praxis. The patient with NIFID was older and exhibited aphasia and dyspraxia. No patient had clinical evidence of motor neurone disease during life, or a mutation in the FUS gene. In the complementary clinical study of 312 patients with clinical syndromes of FTLD, genetic analysis revealed a 6 bp deletion in FUS in 3 patients, of questionable significance. One presented a prototypical picture of FTD, another expressive language disorder, and the third semantic dementia. None showed the early onset age or distinctive 'stereotypic' picture of patients with aFTLD-U. We conclude that aFTLD-U is associated with a distinct clinical form of frontotemporal dementia, potentially allowing identification of such patients in life with a high degree of precision. Whether mutations in the FUS gene cause some cases of FTLD remains unresolved.