Neurodegenerative diseases are characterised by the abnormal filamentous assembly of specific proteins in the central nervous system¹. Human genetic studies established a causal role for protein assembly in neurodegeneration². However, the underlying molecular mechanisms remain largely unknown, which is limiting progress in developing clinical tools for these diseases. Recent advances in electron cryo-microscopy (cryo-EM) have enabled the structures of the protein filaments to be determined from patient brains¹. All diseases studied to date have been characterised by the self-assembly of proteins in homomeric amyloid filaments, including that of TAR DNA-binding protein 43 (TDP-43) in amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration with TDP-43 inclusions (FTLD-TDP) Types A and B^3,4. Here, we used cryo-EM to determine filament structures from the brains of individuals with FTLD-TDP Type C, one of the most common forms of sporadic FTLD-TDP. Unexpectedly, the structures revealed that a second protein, annexin A11 (ANXA11), co-assembles with TDP-43 in heteromeric amyloid filaments. The ordered filament fold is formed by TDP-43 residues G282/284–N345 and ANXA11 residues L39–Y74 from their respective low-complexity domains (LCDs). Regions of TDP-43 and ANXA11 previously implicated in protein-protein interactions form an extensive hydrophobic interface at the centre of the filament fold. Immunoblots of the filaments revealed that the majority of ANXA11 exists as a ~22 kDa N-terminal fragment (NTF) lacking the annexin core domain. Immunohistochemistry of brain sections showed the co-localisation of ANXA11 and TDP-43 in inclusions, redefining the histopathology of FTLD-TDP Type C. This work establishes a central role for ANXA11 in FTLD-TDP Type C. The unprecedented formation of heteromeric amyloid filaments in human brain revises our understanding of amyloid assembly and may be of significance for the pathogenesis of neurodegenerative diseases.