TDP-43 immunoreactive (TDP-43-ir) pathological changes were investigated in the temporal cortex and hippocampus of 11 patients with autosomal dominant familial forms of Alzheimer's disease (FAD), 169 patients with sporadic AD [85 with early onset disease (EOAD) (i.e before 65 years of age), and 84 with late onset after this age (LOAD)], 50 individuals with Down's Syndrome (DS) and 5 patients with primary hippocampal sclerosis (HS). TDP-43-ir pathological changes were present, overall, in 34/180 of AD cases. They were present in 1/11 (9%) FAD, and 9/85 (10%) EOAD patients but were significantly more common (p = 0.003) in LOAD where 24/84 (29%) patients showed such changes. There were no demographic differences, other than onset age, between AD patients with or without TDP-43-ir pathological changes. Double immunolabelling indicated that these TDP-43-ir inclusions were frequently ubiquitinated, but were only rarely AT8 (tau) immunoreactive. Only 3 elderly DS individuals and 4/5 cases of primary HS showed similar changes. Overall, 21.7% of AD cases and 6% DS cases showed hippocampal sclerosis (HS). However, only 9% FAD cases and 16% EOAD cases showed HS, but 29% LOAD cases showed HS. The proportion of EOAD cases with both TDP-43 pathology and HS tended to be greater than those in LOAD, where nearly half of all the cases with TDP-43 pathology did not show HS. The presence of TDP-43-ir changes in AD and DS may therefore be a secondary phenomenon, relating more to ageing than to AD itself. Nevertheless, a challenge to such an interpretation comes from the finding in AD of a strong relationship between TDP-43 pathology and cognitive phenotype. Patients with TDP-43 pathology were significantly more likely to present with an amnestic syndrome than those without (p < 0.0001), in keeping with pathological changes in medial temporal lobe structures. HS was also associated more commonly with an amnestic presentation (p < 0.005), but this association disappeared when TDP-43-positive cases were excluded from the analysis. TDP-43 may, after all, be integral to the pathology of AD, and to some extent determine the clinical phenotype present.