In order to gain insight into the pathogenesis of frontotemporal lobar degeneration (FTLD), the mean tau load in frontal cortex was compared in 34 patients with frontotemporal dementia linked to chromosome 17 (FTDP-17) with 12 different mutations in the tau gene (MAPT), 11 patients with sporadic FTLD with Pick bodies and 25 patients with early onset Alzheimer's disease (EOAD). Tau load was determined, as percentage of tissue occupied by stained product, by image analysis of immunohistochemically stained sections using the phospho-dependent antibodies AT8, AT100 and AT180. With AT8 and AT180 antibodies, the amount of tau was significantly (P < 0.001 in each instance) less than that in EOAD for both FTDP-17 (8.5% and 10.0% respectively) and sporadic FTLD with Pick bodies (16.1% and 10.0% respectively). With AT100, the amount of tau detected in FTDP-17 was 54% (P < 0.001) of that detected in EOAD, but no tau was detected in sporadic FTLD with Pick bodies using this particular antibody. The amount of insoluble tau deposited within the brain in FTDP-17 did not depend in any systematic way upon where the MAPT mutation was topographically located within the gene, or on the physiological or structural change generated by the mutation, regardless of which anti-tau antibody was used. Not only does the amount of tau deposited in the brain differ between the three disorders, but the pattern of phosphorylation of tau also varies according to disease. These findings raise important questions relating to the role of aggregated tau in neurodegeneration - whether this represents an adaptive response which promotes the survival of neurones, or whether it is a detrimental change that directly, or indirectly, brings about the demize of the affected cell.